query stringclasses 259
values | prequel listlengths 0 270 | sequel listlengths 0 162 | during listlengths 0 162 | metadata dict | num_of_frames int64 16 16 | frame_captions listlengths 16 16 | video_id stringlengths 6 45 | frame2video_caption stringlengths 54 2.35k |
|---|---|---|---|---|---|---|---|---|
The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea. | [
"\"Severe earthquake hits Gyeongju, South Korea in 2016\"",
"\"Gyeongju earthquake in South Korea in 2016\"",
"\"Gyeongju earthquake in 2016\"",
"\"2016 Gyeongju earthquake\"",
"The Gyeongju 2016 earthquake"
] | [
"Gyeongju earthquake 2016",
"Rescue people affected by the 2016 Gyeongju earthquake",
"\"Gyeongju earthquake rescue efforts in 2016, South Korea\"",
"\"2016 Gyeongju earthquake\" - related to destruction of buildings",
"Rescue efforts in progress Gyeongju, South Korea in 2016"
] | [
"Gyeongju 2016 earthquake",
"South Korea earthquake 2016"
] | {
"article_url": "https://ko.wikipedia.org/wiki/2016%EB%85%84_%EA%B2%BD%EC%A3%BC_%EC%A7%80%EC%A7%84",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe given query \"The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea\" refers to a specific event. It does not contain any visual data but includes descriptive language that could be used to visualize the event in a video.\n\nEVENTS:\n1. The 2016 Gyeongju earthquake\n2. South Korea",
"en_article_excerpt": "The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.",
"en_article_url": "https://en.wikipedia.org/wiki/2016_Gyeongju_earthquake",
"event_category": "disasters",
"event_name": "gyeongju_earthquake",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThis query could be visualized in a video as a news report or a footage of the event. Given the specifics of the event, it's likely that the video will show destruction and damage, which could be reflected in a caption. Therefore, I will focus on extracting events that could be associated with the video.\n\nEVENTS:\n1. Destruction\n2. Damage\n3. Disaster\n4. Earthquake\n5. Rescue",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nGiven the 2016 Gyeongju earthquake, it is a significant historical event with known visual imagery associated with it. Therefore, it is reasonable to extract events that could be seen in a video related to the earthquake, such as scenes of destruction, people fleeing the area, and rescue efforts.\n\nEVENTS:\n1. Destruction\n2. Rescue\n3. Rescue Efforts\n4. Destruction of Buildings\n5. Rescue Efforts in Progress",
"video_URL": "http://youtube.com/watch?v=t8lY-rjy7-Y",
"video_description": "9월 12일, 경주에서는 역대 최대 규모인 5.8 지진이 일어났다. 지진으로 서울과 수도권까지 전국이 흔들렸고, 이로 인해 전화와 카카오톡도 모두 먹통이었다. 정부는 지진 발생 9분이 지나서야 긴급재난문자를 발송했고 이마저도 서울과 수도권 지역 국민들은 문자를 받지 못했다. #경주 #지진 #재난문자",
"video_id": "t8lY-rjy7-Y"
} | 16 | [
"The image shows a group of people, likely students, rushing down a hallway and up a staircase. They appear to be in a hurry, possibly evacuating the building. The timestamp on the image reads \"7시 44분,\" which indicates the time as 7:44 AM. The setting seems to be a school or similar institution.",
"The image sh... | t8lY-rjy7-Y | A group of students rush down a hallway and up a staircase to evacuate a building in response to a reported earthquake, with a train station scene outside displaying an earthquake detection and the National Assembly reacting to the situation with a press conference. The scene shifts to a warehouse with a person discussing the impact of the earthquake, followed by a bus driving through debris, and a domestic scene showing a person playing with a dog, indicating a domestic reaction to the earthquake. A governmental press conference follows, discussing the situation and the impact on national security services. A screenshot of text messages on a mobile phone further emphasizes the focus on national security services and the operational status of government institutions. The scene shifts to a person addressing the media in a formal setting, apologizing for an incident involving a person, and finally to a chaotic scene in a school hallway where students are reacting to an unexpected event, with students in a classroom displaying confusion and distress. |
The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea. | [
"\"Severe earthquake hits Gyeongju, South Korea in 2016\"",
"\"Gyeongju earthquake in South Korea in 2016\"",
"\"Gyeongju earthquake in 2016\"",
"\"2016 Gyeongju earthquake\"",
"The Gyeongju 2016 earthquake"
] | [
"Gyeongju earthquake 2016",
"Rescue people affected by the 2016 Gyeongju earthquake",
"\"Gyeongju earthquake rescue efforts in 2016, South Korea\"",
"\"2016 Gyeongju earthquake\" - related to destruction of buildings",
"Rescue efforts in progress Gyeongju, South Korea in 2016"
] | [
"Gyeongju 2016 earthquake",
"South Korea earthquake 2016"
] | {
"article_url": "https://ko.wikipedia.org/wiki/2016%EB%85%84_%EA%B2%BD%EC%A3%BC_%EC%A7%80%EC%A7%84",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe given query \"The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea\" refers to a specific event. It does not contain any visual data but includes descriptive language that could be used to visualize the event in a video.\n\nEVENTS:\n1. The 2016 Gyeongju earthquake\n2. South Korea",
"en_article_excerpt": "The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.",
"en_article_url": "https://en.wikipedia.org/wiki/2016_Gyeongju_earthquake",
"event_category": "disasters",
"event_name": "gyeongju_earthquake",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThis query could be visualized in a video as a news report or a footage of the event. Given the specifics of the event, it's likely that the video will show destruction and damage, which could be reflected in a caption. Therefore, I will focus on extracting events that could be associated with the video.\n\nEVENTS:\n1. Destruction\n2. Damage\n3. Disaster\n4. Earthquake\n5. Rescue",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nGiven the 2016 Gyeongju earthquake, it is a significant historical event with known visual imagery associated with it. Therefore, it is reasonable to extract events that could be seen in a video related to the earthquake, such as scenes of destruction, people fleeing the area, and rescue efforts.\n\nEVENTS:\n1. Destruction\n2. Rescue\n3. Rescue Efforts\n4. Destruction of Buildings\n5. Rescue Efforts in Progress",
"video_URL": "http://youtube.com/watch?v=CXe6_CDrdB8",
"video_description": "찐♡기상청사람들 예고 없이 찾아오는 지진! 한 번 발생하면 큰 피해를 주는 지진이기에 감시하는 일은 한시도 소홀히 할 수 없습니다. 그 최전방에서... 1년 365일 24시간 지진화산 감시 및 분석 업무에 최선을 다해 신속·정확하게 알리고자 노력하는 사람이 있습니다. 지진화산 감시 및 분석 업무 9년 차... 반복되는 주간 근무와 야간 근무가 힘들긴 하지만, 누구보다 묵묵하게 자리를 지키고 있는 진득한 뚝배기 같은 지진업무의 대들보!! 지진화산 감시 업무를 담당하는 기상청 사람을 만나고 왔습니다.(^_^)v 우리나라에서 가장 강력했던 규모 5.8의 경주 지진 때 지진감시 및 분석을 수행했던 기상청 사람은 누구? 지진감시 업무를 들어올 때마다 긴장되는 기상청 사람은 누구? 국민의 안전을 누구보다 걱정하는 기상청 사람은 누구? 지진 올바로 알고 지진 발생 시 빠르게 대처하길 바라는 기상청 사람은 누구? 최근 지진 안전을 홍보해 주시는 지진희 배우님을 그 누구보다 고마워하는 기상청 사람은 누구? 만나볼까요 :) #기상청사람들 #지진화산 #행동하는파수꾼",
"video_id": "CXe6_CDrdB8"
} | 16 | [
"In this frame, a man is standing in front of a wall of multiple monitors displaying various data and graphs. He is wearing a blue shirt and a white watch, and appears to be presenting or discussing the content on the screens. The text on the screen reads \"Q 엄무 소개?\" which translates to \"Q Introduction to work?\"... | CXe6_CDrdB8 | A major earthquake occurred in Gyeongju, South Korea with a magnitude of 5.8. The event was the largest to have occurred in Korea, and it was detected through multiple monitoring stations. The earthquake was a significant event for the region, and data analysis was crucial in determining its location and intensity. The data analysis involved experts in monitoring stations who interpreted the seismic waves to provide timely warnings to the public. |
The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea. | [
"\"Severe earthquake hits Gyeongju, South Korea in 2016\"",
"\"Gyeongju earthquake in South Korea in 2016\"",
"\"Gyeongju earthquake in 2016\"",
"\"2016 Gyeongju earthquake\"",
"The Gyeongju 2016 earthquake"
] | [
"Gyeongju earthquake 2016",
"Rescue people affected by the 2016 Gyeongju earthquake",
"\"Gyeongju earthquake rescue efforts in 2016, South Korea\"",
"\"2016 Gyeongju earthquake\" - related to destruction of buildings",
"Rescue efforts in progress Gyeongju, South Korea in 2016"
] | [
"Gyeongju 2016 earthquake",
"South Korea earthquake 2016"
] | {
"article_url": "https://ko.wikipedia.org/wiki/2016%EB%85%84_%EA%B2%BD%EC%A3%BC_%EC%A7%80%EC%A7%84",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe given query \"The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea\" refers to a specific event. It does not contain any visual data but includes descriptive language that could be used to visualize the event in a video.\n\nEVENTS:\n1. The 2016 Gyeongju earthquake\n2. South Korea",
"en_article_excerpt": "The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.",
"en_article_url": "https://en.wikipedia.org/wiki/2016_Gyeongju_earthquake",
"event_category": "disasters",
"event_name": "gyeongju_earthquake",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThis query could be visualized in a video as a news report or a footage of the event. Given the specifics of the event, it's likely that the video will show destruction and damage, which could be reflected in a caption. Therefore, I will focus on extracting events that could be associated with the video.\n\nEVENTS:\n1. Destruction\n2. Damage\n3. Disaster\n4. Earthquake\n5. Rescue",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nGiven the 2016 Gyeongju earthquake, it is a significant historical event with known visual imagery associated with it. Therefore, it is reasonable to extract events that could be seen in a video related to the earthquake, such as scenes of destruction, people fleeing the area, and rescue efforts.\n\nEVENTS:\n1. Destruction\n2. Rescue\n3. Rescue Efforts\n4. Destruction of Buildings\n5. Rescue Efforts in Progress",
"video_URL": "http://youtube.com/watch?v=uF3x63vtddc",
"video_description": "\"경주 지진, 지하수는 1~2일 전에 미리 알았다\" 지난 12일과 19일 경주에서 규모 5.8을 비롯한 대형 지진이 잇따라 발생하기 하루 이틀 전 주변 지하수 수위가 크게 상승해, 향후 지진예측에 활용할 수 있다는 주장이 나왔습니다. 정상용 부경대 지구환경과학과 교수는 진앙에서 가장 가까운 경북 경주시 산내면에 있는 지하수 관측소 자료를 분석한 결과, 이 같이 밝혔습니다. 자료를 보면 지난 11일 암반 지하수 수위는 하루 사이에 무려 49㎝ 상승했고, 17일에도 39㎝나 올라갔습니다. 정 교수는 지진에 앞서 암석에 강한 압력이 가해져 지하수 수위가 많이 올라간 것으로 보인다며, 규모 4.5 이상의 지진에만 영향을 받았다고 설명했습니다. 연합뉴스TV : 02-398-4409(제보) 4441(기사문의), 카톡/라인 jebo23 (끝)",
"video_id": "uF3x63vtddc"
} | 16 | [
"In the image, two news anchors are standing in a news studio. The male anchor is on the left and the female anchor is on the right. They appear to be delivering a news report. The studio background shows multiple screens displaying various images, likely from other news stories or related footage. \n\nThe text on ... | uF3x63vtddc | In the given video, it appears to be documenting the story of an earthquake in Gyeongju, South Korea, which occurred on 1-2 days before the time of 09:34. The footage focuses on a news report that highlighted the early warning signs of the earthquake, including the indication of groundwater being affected. The incident was reportedly related to a machine in Gyeongju that was operational for 1-2 days before the event. The footage shows a series of images and descriptions of the aftermath of the earthquake, including damage to buildings, debris, and flooding. The narrative also includes news segments and footage from news channels, providing context to the incident. |
The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea. | [
"\"Severe earthquake hits Gyeongju, South Korea in 2016\"",
"\"Gyeongju earthquake in South Korea in 2016\"",
"\"Gyeongju earthquake in 2016\"",
"\"2016 Gyeongju earthquake\"",
"The Gyeongju 2016 earthquake"
] | [
"Gyeongju earthquake 2016",
"Rescue people affected by the 2016 Gyeongju earthquake",
"\"Gyeongju earthquake rescue efforts in 2016, South Korea\"",
"\"2016 Gyeongju earthquake\" - related to destruction of buildings",
"Rescue efforts in progress Gyeongju, South Korea in 2016"
] | [
"Gyeongju 2016 earthquake",
"South Korea earthquake 2016"
] | {
"article_url": "https://ko.wikipedia.org/wiki/2016%EB%85%84_%EA%B2%BD%EC%A3%BC_%EC%A7%80%EC%A7%84",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe given query \"The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea\" refers to a specific event. It does not contain any visual data but includes descriptive language that could be used to visualize the event in a video.\n\nEVENTS:\n1. The 2016 Gyeongju earthquake\n2. South Korea",
"en_article_excerpt": "The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.",
"en_article_url": "https://en.wikipedia.org/wiki/2016_Gyeongju_earthquake",
"event_category": "disasters",
"event_name": "gyeongju_earthquake",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThis query could be visualized in a video as a news report or a footage of the event. Given the specifics of the event, it's likely that the video will show destruction and damage, which could be reflected in a caption. Therefore, I will focus on extracting events that could be associated with the video.\n\nEVENTS:\n1. Destruction\n2. Damage\n3. Disaster\n4. Earthquake\n5. Rescue",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nGiven the 2016 Gyeongju earthquake, it is a significant historical event with known visual imagery associated with it. Therefore, it is reasonable to extract events that could be seen in a video related to the earthquake, such as scenes of destruction, people fleeing the area, and rescue efforts.\n\nEVENTS:\n1. Destruction\n2. Rescue\n3. Rescue Efforts\n4. Destruction of Buildings\n5. Rescue Efforts in Progress",
"video_URL": "http://youtube.com/watch?v=nln3QFJggKs",
"video_description": "106억 → 58억…경주 지진 피해금액 반토막 잇단 강진으로 100억원이 넘는 것으로 추정됐던 경주시의 공식 피해규모가 58억원 수준에 그치는 것으로 집계됐습니다. 경주시가 지진 발생 뒤 전수조사를 통해 발표한 잠정 추산 피해액은 약 107억원이었지만 사유재산 피해는 대부분 공식 피해액에서 빠졌습니다. 규정상 기와가 떨어지거나 벽체에 금이 가고 담장이 파손된 것은 재산피해 산정되지 않습니다. 이에 따라 사유재산 중에는 지진으로 주택이 완전히 부서지거나 반쯤 허물어진 것만 공식 피해액에 들어갔습니다. 연합뉴스TV : 02-398-4409(제보) 4441(기사문의), 카톡/라인 jebo23 (끝)",
"video_id": "nln3QFJggKs"
} | 16 | [
"The image shows a news broadcast from Yonhap News TV, featuring two news anchors. The male anchor on the left is dressed in a dark suit and red tie, while the female anchor on the right wears a blue blazer over a pink top. Both are standing in front of a backdrop depicting a nighttime street scene. There is a larg... | nln3QFJggKs | The video is a series of images depicting the aftermath of a massive earthquake in Gyeongju, South Korea, where a 4.5 aftershock has led to significant reductions in damage estimates. The video focuses on:
- A 6-frame series of images showing the aftermath of the earthquake, highlighting the extent of damage to roads, buildings, and infrastructure.
- The damage includes cracked or destroyed buildings, roads, and debris.
- The estimated damage amounts have been reduced from 106 billion to 58 billion won.
- The scene emphasizes the community's efforts to recover and rebuild, with people gathering for aid and assistance.
- The video captures the ongoing process of inspection and repair, highlighting the urgency of addressing the earthquake's impact.
The scenes are set in a predominantly somber and quiet atmosphere, emphasizing the extent of the destruction and the need for immediate attention to prevent further deterioration. |
The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea. | [
"\"Severe earthquake hits Gyeongju, South Korea in 2016\"",
"\"Gyeongju earthquake in South Korea in 2016\"",
"\"Gyeongju earthquake in 2016\"",
"\"2016 Gyeongju earthquake\"",
"The Gyeongju 2016 earthquake"
] | [
"Gyeongju earthquake 2016",
"Rescue people affected by the 2016 Gyeongju earthquake",
"\"Gyeongju earthquake rescue efforts in 2016, South Korea\"",
"\"2016 Gyeongju earthquake\" - related to destruction of buildings",
"Rescue efforts in progress Gyeongju, South Korea in 2016"
] | [
"Gyeongju 2016 earthquake",
"South Korea earthquake 2016"
] | {
"article_url": "https://ko.wikipedia.org/wiki/2016%EB%85%84_%EA%B2%BD%EC%A3%BC_%EC%A7%80%EC%A7%84",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe given query \"The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea\" refers to a specific event. It does not contain any visual data but includes descriptive language that could be used to visualize the event in a video.\n\nEVENTS:\n1. The 2016 Gyeongju earthquake\n2. South Korea",
"en_article_excerpt": "The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.",
"en_article_url": "https://en.wikipedia.org/wiki/2016_Gyeongju_earthquake",
"event_category": "disasters",
"event_name": "gyeongju_earthquake",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThis query could be visualized in a video as a news report or a footage of the event. Given the specifics of the event, it's likely that the video will show destruction and damage, which could be reflected in a caption. Therefore, I will focus on extracting events that could be associated with the video.\n\nEVENTS:\n1. Destruction\n2. Damage\n3. Disaster\n4. Earthquake\n5. Rescue",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nGiven the 2016 Gyeongju earthquake, it is a significant historical event with known visual imagery associated with it. Therefore, it is reasonable to extract events that could be seen in a video related to the earthquake, such as scenes of destruction, people fleeing the area, and rescue efforts.\n\nEVENTS:\n1. Destruction\n2. Rescue\n3. Rescue Efforts\n4. Destruction of Buildings\n5. Rescue Efforts in Progress",
"video_URL": "http://youtube.com/watch?v=n6UmQ9Kw2GQ",
"video_description": "경주 지진 이후 여진 307회 발생…\"강도 약해져\" 지난 12일 저녁 발생한 규모 5.8의 경주 지진 탓에 여진이 307차례 발생한 것으로 집계됐습니다. 기상청에 따르면 오늘 오후 4시까지 경주 지진 이후 여진은 규모 1.5∼3.0도 292회 등 모두 307차례 발생했습니다. 기상청 관계자는 \"경주 지진 탓에 여진이 계속 발생하고 있지만 강도가 약해지고 있다\"며 \"앞으로 일어날 여진 규모는 3.0 이하로 예상되기 때문에 그렇게 우려하지 않아도 될 것 같다\"고 말했습니다. 연합뉴스TV : 02-398-4409(제보) 4441(기사문의), 카톡/라인 jebo23 (끝)",
"video_id": "n6UmQ9Kw2GQ"
} | 16 | [
"The image shows a news broadcast with two news anchors wearing traditional Korean clothing. The screen behind them displays a weather map highlighting a region in South Korea, indicating a severe weather event. The text on the screen reads, \"경주 지진 이후 여진 307회 발생... '강도 약해져'\" which translates to \"Following the Gy... | n6UmQ9Kw2GQ | The video depicts a severe weather event and aftershocks in South Korea, specifically following the Gyeongju earthquake. The narrative emphasizes the aftermath of the earthquake and its ongoing impact on the local community, particularly women. The video focuses on various aspects of their lives, including their daily routines, experiences, and perceptions of the event. The narrative highlights the reduction in intensity of the earthquake over time, suggesting that the community has adapted to the ongoing challenges. The video also mentions the presence of various household items and logistics services, such as delivery times and household essentials. |
The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea. | [
"\"Severe earthquake hits Gyeongju, South Korea in 2016\"",
"\"Gyeongju earthquake in South Korea in 2016\"",
"\"Gyeongju earthquake in 2016\"",
"\"2016 Gyeongju earthquake\"",
"The Gyeongju 2016 earthquake"
] | [
"Gyeongju earthquake 2016",
"Rescue people affected by the 2016 Gyeongju earthquake",
"\"Gyeongju earthquake rescue efforts in 2016, South Korea\"",
"\"2016 Gyeongju earthquake\" - related to destruction of buildings",
"Rescue efforts in progress Gyeongju, South Korea in 2016"
] | [
"Gyeongju 2016 earthquake",
"South Korea earthquake 2016"
] | {
"article_url": "https://ko.wikipedia.org/wiki/2016%EB%85%84_%EA%B2%BD%EC%A3%BC_%EC%A7%80%EC%A7%84",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe given query \"The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea\" refers to a specific event. It does not contain any visual data but includes descriptive language that could be used to visualize the event in a video.\n\nEVENTS:\n1. The 2016 Gyeongju earthquake\n2. South Korea",
"en_article_excerpt": "The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.",
"en_article_url": "https://en.wikipedia.org/wiki/2016_Gyeongju_earthquake",
"event_category": "disasters",
"event_name": "gyeongju_earthquake",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThis query could be visualized in a video as a news report or a footage of the event. Given the specifics of the event, it's likely that the video will show destruction and damage, which could be reflected in a caption. Therefore, I will focus on extracting events that could be associated with the video.\n\nEVENTS:\n1. Destruction\n2. Damage\n3. Disaster\n4. Earthquake\n5. Rescue",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nGiven the 2016 Gyeongju earthquake, it is a significant historical event with known visual imagery associated with it. Therefore, it is reasonable to extract events that could be seen in a video related to the earthquake, such as scenes of destruction, people fleeing the area, and rescue efforts.\n\nEVENTS:\n1. Destruction\n2. Rescue\n3. Rescue Efforts\n4. Destruction of Buildings\n5. Rescue Efforts in Progress",
"video_URL": "http://youtube.com/watch?v=uoagwCVoHGM",
"video_description": "경주 지진으로 8명 부상…재산피해 253건 어제 저녁 경주 일대에서 발생한 지진으로 인해 8명이 다치고 카카오톡과 국민안전처 홈페이지 등 통신망이 일시적으로 마비되는 등의 장애가 있었습니다. 국민안전처는 경주 일대에서 오늘과 어제에 걸쳐 일어난 지진으로 인해 80살 여성 황 모 씨 등 8명이 경상을 입고 253건의 재산피해가 신고됐다고 밝혔습니다. 이번 지진으로 인해 KTX 대전~부산 구간이 5분가량 정차하기도 했습니다. 연합뉴스TV : 02-398-4409(제보) 4441(기사문의), 카톡/라인 jebo23 (끝)",
"video_id": "uoagwCVoHGM"
} | 16 | [
"A news anchor in a purple blazer and light pink top is presenting a story. Behind her is a black-and-white surveillance image showing a scene with water, possibly from a flood, as it flows over a barrier. The image features several potted plants and a railing on the left side. The text on the screen, in Korean, me... | uoagwCVoHGM | The video depicts the aftermath of a significant earthquake in Gyeongju, South Korea, resulting in 8 injuries and 253 property damages. The event occurred after a 6.8 magnitude earthquake hit the city, causing widespread destruction and disruption. The visual details show various scenes of damaged buildings, debris, and injured people. The footage includes footage of emergency responders, cleanup efforts, and a bustling airport scene. The overall context suggests that the earthquake has had a severe impact on the city, including structural damage to buildings and significant loss of life. |
The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea. | [
"\"Severe earthquake hits Gyeongju, South Korea in 2016\"",
"\"Gyeongju earthquake in South Korea in 2016\"",
"\"Gyeongju earthquake in 2016\"",
"\"2016 Gyeongju earthquake\"",
"The Gyeongju 2016 earthquake"
] | [
"Gyeongju earthquake 2016",
"Rescue people affected by the 2016 Gyeongju earthquake",
"\"Gyeongju earthquake rescue efforts in 2016, South Korea\"",
"\"2016 Gyeongju earthquake\" - related to destruction of buildings",
"Rescue efforts in progress Gyeongju, South Korea in 2016"
] | [
"Gyeongju 2016 earthquake",
"South Korea earthquake 2016"
] | {
"article_url": "https://ko.wikipedia.org/wiki/2016%EB%85%84_%EA%B2%BD%EC%A3%BC_%EC%A7%80%EC%A7%84",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe given query \"The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea\" refers to a specific event. It does not contain any visual data but includes descriptive language that could be used to visualize the event in a video.\n\nEVENTS:\n1. The 2016 Gyeongju earthquake\n2. South Korea",
"en_article_excerpt": "The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.",
"en_article_url": "https://en.wikipedia.org/wiki/2016_Gyeongju_earthquake",
"event_category": "disasters",
"event_name": "gyeongju_earthquake",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThis query could be visualized in a video as a news report or a footage of the event. Given the specifics of the event, it's likely that the video will show destruction and damage, which could be reflected in a caption. Therefore, I will focus on extracting events that could be associated with the video.\n\nEVENTS:\n1. Destruction\n2. Damage\n3. Disaster\n4. Earthquake\n5. Rescue",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nGiven the 2016 Gyeongju earthquake, it is a significant historical event with known visual imagery associated with it. Therefore, it is reasonable to extract events that could be seen in a video related to the earthquake, such as scenes of destruction, people fleeing the area, and rescue efforts.\n\nEVENTS:\n1. Destruction\n2. Rescue\n3. Rescue Efforts\n4. Destruction of Buildings\n5. Rescue Efforts in Progress",
"video_URL": "http://youtube.com/watch?v=5p1N4MtudiI",
"video_description": "경주 지진 316회 이어가다 현재는 '잠잠' 지난 12일 발생한 규모 5.8의 경주 지진의 여진이 316회 발생한 것으로 집계됐습니다. 기상청은 오늘 오전 11시 기준 여진 횟수가 316회라고 발표했습니다. 규모별 횟수는 1.5에서 3.0이 301차례, 3.0에서 4.0은 14차례, 4.0에서 5.0은 1차례 감지됐습니다. 사흘째 계속된 여진은 일단 멈추는 징후를 보이고 있는 것으로 기상청은 파악하고 있습니다. 연합뉴스TV : 02-398-4409(제보) 4441(기사문의), 카톡/라인 jebo23 (끝)",
"video_id": "5p1N4MtudiI"
} | 16 | [
"The image is a still from a news broadcast on EBS TV. Two presenters are standing in front of a cityscape backdrop, featuring a bridge over a river with buildings in the background. The presenters are dressed in traditional Korean hanbok attire. The text at the bottom of the screen in Korean mentions \"The 316th I... | 5p1N4MtudiI | The 316th Invention Exhibition, a notable event in Korea, has been declared "Deadlock" after 31 years without any notable event. The event features a backdrop of traditional Korean attire and is centered on the 316th Invention Exhibition, a showcase of Korean inventions. The exhibition has been marked by severe weather patterns and ongoing seismic activity.
The event is currently surrounded by a news broadcast that highlights the status as "Deadlock," indicating that the weather has worsened and seismic activity continues to rise. Seismic data, including maps, radar images, and radar waveforms, are presented, suggesting that the situation remains critical. The event continues to attract significant public attention with the number 89 being displayed prominently on the screen, possibly representing an ongoing analysis of seismic events. |
The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea. | [
"\"Severe earthquake hits Gyeongju, South Korea in 2016\"",
"\"Gyeongju earthquake in South Korea in 2016\"",
"\"Gyeongju earthquake in 2016\"",
"\"2016 Gyeongju earthquake\"",
"The Gyeongju 2016 earthquake"
] | [
"Gyeongju earthquake 2016",
"Rescue people affected by the 2016 Gyeongju earthquake",
"\"Gyeongju earthquake rescue efforts in 2016, South Korea\"",
"\"2016 Gyeongju earthquake\" - related to destruction of buildings",
"Rescue efforts in progress Gyeongju, South Korea in 2016"
] | [
"Gyeongju 2016 earthquake",
"South Korea earthquake 2016"
] | {
"article_url": "https://ko.wikipedia.org/wiki/2016%EB%85%84_%EA%B2%BD%EC%A3%BC_%EC%A7%80%EC%A7%84",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe given query \"The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea\" refers to a specific event. It does not contain any visual data but includes descriptive language that could be used to visualize the event in a video.\n\nEVENTS:\n1. The 2016 Gyeongju earthquake\n2. South Korea",
"en_article_excerpt": "The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.",
"en_article_url": "https://en.wikipedia.org/wiki/2016_Gyeongju_earthquake",
"event_category": "disasters",
"event_name": "gyeongju_earthquake",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThis query could be visualized in a video as a news report or a footage of the event. Given the specifics of the event, it's likely that the video will show destruction and damage, which could be reflected in a caption. Therefore, I will focus on extracting events that could be associated with the video.\n\nEVENTS:\n1. Destruction\n2. Damage\n3. Disaster\n4. Earthquake\n5. Rescue",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nGiven the 2016 Gyeongju earthquake, it is a significant historical event with known visual imagery associated with it. Therefore, it is reasonable to extract events that could be seen in a video related to the earthquake, such as scenes of destruction, people fleeing the area, and rescue efforts.\n\nEVENTS:\n1. Destruction\n2. Rescue\n3. Rescue Efforts\n4. Destruction of Buildings\n5. Rescue Efforts in Progress",
"video_URL": "http://youtube.com/watch?v=roaxxb8zRLU",
"video_description": "국제신문 뉴스&이슈 구독하기 http://goo.gl/FFQL4 국제신문 근교산&기획특집 구독하기 http://goo.gl/SM0fM 국제신문 기사보기 http://iii.im/pk3Z 국제신문 영상팀이 경주 지진피해 복구 현장을 드론으로 담았다.",
"video_id": "roaxxb8zRLU"
} | 16 | [
"The image shows an aerial view of a traditional Korean village with numerous houses featuring black-tiled roofs. The text on the image is in Korean and reads \"하늘에서 본 경주\" (Jeju seen from the sky) and \"9.24일 도요미 경주시 황남동 일대\" (On September 24, Toyomi, Hwangnam-dong area of Gyeongju City). The view captures the den... | roaxxb8zRLU | The video depicts a traditional Korean village, known as Gyeongju, with numerous black-tiled roofs, amidst the scenic landscape of the Korean countryside. The village is the focus of the video, showcasing various scenes of daily life, architectural structures, and maintenance work. The restoration of historical buildings and homes in the village is a significant aspect of the video, highlighting the efforts of the community to preserve cultural heritage. The video transitions from a serene and unobstructed view of the village to a focus on restoration work, showcasing the community's dedication to preserving the region's history and cultural significance.
The video also highlights various aspects of traditional Korean culture, including traditional architecture, festivals, and community activities. The inclusion of natural elements, such as greenery and mountains, emphasizes the natural beauty of the region. The video provides a visual and informational overview of Gyeongju, showcasing its unique cultural heritage and significance.
Throughout the video, the narrative is delivered through the descriptions of individual frames, each focusing on specific aspects of the village, such as architecture, maintenance work, and community activities. The videos transition smoothly, seamlessly moving between these different perspectives, providing a comprehensive and engaging visual narrative of the village and its significance. |
The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea. | [
"\"Severe earthquake hits Gyeongju, South Korea in 2016\"",
"\"Gyeongju earthquake in South Korea in 2016\"",
"\"Gyeongju earthquake in 2016\"",
"\"2016 Gyeongju earthquake\"",
"The Gyeongju 2016 earthquake"
] | [
"Gyeongju earthquake 2016",
"Rescue people affected by the 2016 Gyeongju earthquake",
"\"Gyeongju earthquake rescue efforts in 2016, South Korea\"",
"\"2016 Gyeongju earthquake\" - related to destruction of buildings",
"Rescue efforts in progress Gyeongju, South Korea in 2016"
] | [
"Gyeongju 2016 earthquake",
"South Korea earthquake 2016"
] | {
"article_url": "https://ko.wikipedia.org/wiki/2016%EB%85%84_%EA%B2%BD%EC%A3%BC_%EC%A7%80%EC%A7%84",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe given query \"The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea\" refers to a specific event. It does not contain any visual data but includes descriptive language that could be used to visualize the event in a video.\n\nEVENTS:\n1. The 2016 Gyeongju earthquake\n2. South Korea",
"en_article_excerpt": "The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.",
"en_article_url": "https://en.wikipedia.org/wiki/2016_Gyeongju_earthquake",
"event_category": "disasters",
"event_name": "gyeongju_earthquake",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThis query could be visualized in a video as a news report or a footage of the event. Given the specifics of the event, it's likely that the video will show destruction and damage, which could be reflected in a caption. Therefore, I will focus on extracting events that could be associated with the video.\n\nEVENTS:\n1. Destruction\n2. Damage\n3. Disaster\n4. Earthquake\n5. Rescue",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nGiven the 2016 Gyeongju earthquake, it is a significant historical event with known visual imagery associated with it. Therefore, it is reasonable to extract events that could be seen in a video related to the earthquake, such as scenes of destruction, people fleeing the area, and rescue efforts.\n\nEVENTS:\n1. Destruction\n2. Rescue\n3. Rescue Efforts\n4. Destruction of Buildings\n5. Rescue Efforts in Progress",
"video_URL": "http://youtube.com/watch?v=3ojHTwrvx90",
"video_description": "박 대통령, 경주 지진현장 방문 \"특별재난지역 선포\" 박근혜 대통령은 최근 강진이 잇따라 발생한 경북 경주 지역을 찾아 특별재난지역 선포를 적극 검토하라고 지시했습니다. 이어 지진 등 재난 안전 시스템을 원점에서 재점검할 것도 주문했습니다. 박 대통령은 아울러 인근 월성 원자력발전소를 찾아 \"원전은 단 한치의 실수도 용납되지 않는다\"며 지진 방재대책을 철저히 재점검해달라고 당부했습니다. 박 대통령의 지시에 국민안전처 또한 경주시를 특별재난지역으로 선포할 예정입니다. 특별재난지역으로 지정되면 지자체의 피해 복구비 부담이 완화되고, 피해 주민들은 공공요금 감면혜택을 받게됩니다. 연합뉴스TV : 02-398-4409(제보) 4441(기사문의), 카톡/라인 jebo23 (끝)",
"video_id": "3ojHTwrvx90"
} | 16 | [
"The current frame features two news anchors presenting a news story. In the background, a group of people is gathered, with one woman in a yellow jacket speaking to the crowd. The text on the screen indicates the news is about President Park visiting a local community, mentioning a special voting area.",
"In the... | 3ojHTwrvx90 | President Park's delegation, consisting of officials and staff in yellow jackets, is actively involved in the voting area for a special election in Jeonju. They appear to be overseeing the smooth operation of the voting process, ensuring its efficient management amidst challenges posed by the damaged building. |
The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea. | [
"\"Severe earthquake hits Gyeongju, South Korea in 2016\"",
"\"Gyeongju earthquake in South Korea in 2016\"",
"\"Gyeongju earthquake in 2016\"",
"\"2016 Gyeongju earthquake\"",
"The Gyeongju 2016 earthquake"
] | [
"Gyeongju earthquake 2016",
"Rescue people affected by the 2016 Gyeongju earthquake",
"\"Gyeongju earthquake rescue efforts in 2016, South Korea\"",
"\"2016 Gyeongju earthquake\" - related to destruction of buildings",
"Rescue efforts in progress Gyeongju, South Korea in 2016"
] | [
"Gyeongju 2016 earthquake",
"South Korea earthquake 2016"
] | {
"article_url": "https://ko.wikipedia.org/wiki/2016%EB%85%84_%EA%B2%BD%EC%A3%BC_%EC%A7%80%EC%A7%84",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe given query \"The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea\" refers to a specific event. It does not contain any visual data but includes descriptive language that could be used to visualize the event in a video.\n\nEVENTS:\n1. The 2016 Gyeongju earthquake\n2. South Korea",
"en_article_excerpt": "The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.",
"en_article_url": "https://en.wikipedia.org/wiki/2016_Gyeongju_earthquake",
"event_category": "disasters",
"event_name": "gyeongju_earthquake",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThis query could be visualized in a video as a news report or a footage of the event. Given the specifics of the event, it's likely that the video will show destruction and damage, which could be reflected in a caption. Therefore, I will focus on extracting events that could be associated with the video.\n\nEVENTS:\n1. Destruction\n2. Damage\n3. Disaster\n4. Earthquake\n5. Rescue",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2016 Gyeongju earthquake occurred on September 12 near Gyeongju, South Korea.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nGiven the 2016 Gyeongju earthquake, it is a significant historical event with known visual imagery associated with it. Therefore, it is reasonable to extract events that could be seen in a video related to the earthquake, such as scenes of destruction, people fleeing the area, and rescue efforts.\n\nEVENTS:\n1. Destruction\n2. Rescue\n3. Rescue Efforts\n4. Destruction of Buildings\n5. Rescue Efforts in Progress",
"video_URL": "http://youtube.com/watch?v=w6jdwGYcKZU",
"video_description": "안전처 \"경주지진 부상자 23명ㆍ재산피해 1,118건\" 국민안전처는 경주에서 발생한 규모 5.8의 지진으로 부상자 23명이 발생하고, 재산상 피해는 1천118건으로 집계됐다고 밝혔습니다. 안전처와 보건복지부에 따르면 부상자 23명 가운데 7명은 병원에서 치료를 받고 귀가했으며, 16명은 입원 중입니다. 지역별 부상자는 경북이 12명으로 가장 많고, 울산 4명, 대구 2명, 부산·인천·충북·전남·경남 각 1명입니다. 안전처는 균열이 발생한 주택을 대상으로 구조기술사 등과 함께 안전점검을 하고 귀가 여부와 추가 안전조치 등을 검토할 예정입니다. 연합뉴스TV : 02-398-4409(제보) 4441(기사문의), 카톡/라인 jebo23 (끝)",
"video_id": "w6jdwGYcKZU"
} | 16 | [
"A news anchor is delivering a report in front of a cityscape background. The text on the screen indicates a news story about a collision involving a government official, mentioning \"collision\" and \"government official\" along with details about the incident. The temperature and time are also displayed.",
"The... | w6jdwGYcKZU | The video depicts a single news report about an earthquake that occurred in Gyeongju, South Korea. The event began with a collision involving a government official and resulted in 23 injuries and 1,118 property damage cases. The initial scenes show the aftermath of the earthquake, including a snowy and rainy scene, with the government official being treated at a hospital. The footage then shifts to a series of close-ups of the damaged area, including cracked walls, broken furniture, and debris. The earthquake and typhoon ( typhoon "12-1-15") mentioned in the video indicate that the earthquake was one of several devastating natural disasters in South Korea. The damaged areas are shown in the early morning hours, with the focus on the aftermath of the earthquakes and the ongoing efforts to respond to the disaster. |
The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017. | [
"\"2017 aftershocks in Pohang, South Korea, North Gyeongsang Province\"",
"' aftershocks and aftershocks in November 15, 2017, in Pohang, North Gyeongsang Province, South Korea'",
"Aftershock in November in Pohang, South Korea",
"\"Natural disasters in November 15, 2018 in Pohang, North Gyeongsang Province, S... | [
"\"earthquake in Pohang, South Korea, North Gyeongsang Province on 2017, exactly 15 days ago\"",
"\"earthquake in November 15, 2017 at 15:00 in Pohang, North Gyeongsang Province, South Korea\"",
"\"North Korean earthquake in November in South Korea\"",
"\"November 2017 earthquake in Pohang, South Korea\"",
... | [
"\"protesters in Pohang, South Korea, North Gyeongsang Province during 2017 Pohang earthquake\"",
"\"Protesters march in the streets of Pohang, South Korea on November 15, 2017 during a protest.\"",
"Protesters marching in the streets of Pohang, South Korea, North Gyeongsang Province, in November",
"\"Protest... | {
"article_url": "https://ko.wikipedia.org/wiki/2017%EB%85%84_%ED%8F%AC%ED%95%AD_%EC%A7%80%EC%A7%84",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe given query is to extract events that could be visualized in a video and are likely to be related to the 2017 Pohang earthquake. We should consider events that could be seen in a video of the event.\n\nGiven the nature of the query, it's likely that the event could be visualized in a video if it involves significant damage or destruction, people affected by the earthquake, or related to the event itself. The event that makes sense to extract is \"Protesters marching in the streets\" because it is a plausible and concrete event that could be seen in a video of the event. This event is related to the query, which is about the 2017 Pohang earthquake. \n\nEVENTS:\n1. Protesters marching in the streets\n2. People affected by the earthquake\n3. Damage to structures\n4. People injured\n5. Rescue efforts",
"en_article_excerpt": "The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.",
"en_article_url": "https://en.wikipedia.org/wiki/2017_Pohang_earthquake",
"event_category": "disasters",
"event_name": "pohang_earthquake",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThe 2017 Pohang earthquake is a strong seismic event that occurred in South Korea. Given the nature of the query, it is reasonable to extract events that are related to natural disasters, particularly those that are publicly documented and visually verifiable. I will consider the query as a whole and extract events that are likely to be visualized in a video, such as news reports, images, or footage of the aftermath of the earthquake.\n\nEVENTS:\n1. Aftershocks\n2. Emergency response efforts\n3. Rescue operations\n4. Destruction and damage to buildings\n5. Evacuation efforts",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\n\nConsidering the query contains specific geospatial information (Korea), locations (Heunghae and Pohang), and time (November 15, 2017), it suggests that the event might be a natural disaster or a disaster that occurred in this location. Given the location and date, it is likely that the event could be visualized in a video related to natural disasters or emergencies.\n\nEVENTS:\n\n1. Earthquake\n2. Landslide\n3. Tsunami",
"video_URL": "http://youtube.com/watch?v=oK2O8Xtkrvs",
"video_description": "15일 오후 2시 29분쯤 경북 포항시 북구 북쪽 9㎞ 지역에서 규모 5.4의 지진이 발생했다. 오마이TV는 지진 발생 당시 포항시 북구 한살림협동조합 매장을 촬영한 CCTV영상을 제보받아 공개한다. 이 영상에는 갑자기 진열대 물건들이 쏟아져내리는 장면이 들어있다. 한살림협동조합에 근무하는 박혜령(경북 영덕군)씨는 영상을 제보하면서 \"당시 사무실에 있었는데 벽지와 창문이 20초 이상 심하게 흔들렸다\"며 \"아직도 가슴이 진정되지 않는다\"고 말했다. 그는 \"바닥이 계속 흔들려 모두 밖으로 나왔고 안으로 들어갈 엄두를 내지 못했다\"며 \"이후 여진을 알리는 재난문자가 세 차례 왔다\"고 당시 상황을 설명했다. (영상제보 : 박혜령 씨/ 영상편집 : 황지희 기자)",
"video_id": "oK2O8Xtkrvs"
} | 16 | [
"The image shows a store interior with shelves stocked with various products. A person is bending over in the middle aisle, appearing to look for something or reach for an item on a lower shelf. The text on the image indicates that this is a scene from CCTV footage related to a shoplifting incident, as it mentions ... | oK2O8Xtkrvs | A video contains footage of a person attempting to shoplift in a grocery store. The scene unfolds with the person bending over in the middle aisle, inspecting products and then grabbing items, which is captured by CCTV footage. The action is repeated throughout the video, with the person gathering items and moving to a different aisle. The store is fully stocked with products, and the action is in the midst of a shoplifting incident. The incident is caught on camera, and the person's actions are distressing and chaotic. The incident is also reported as a crime, with a caption mentioning the Pohang earthquake. The video continues to document the person's actions and the aftermath, including a transition to an informational segment featuring the OhmyTV logo and website. |
The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017. | [
"\"2017 aftershocks in Pohang, South Korea, North Gyeongsang Province\"",
"' aftershocks and aftershocks in November 15, 2017, in Pohang, North Gyeongsang Province, South Korea'",
"Aftershock in November in Pohang, South Korea",
"\"Natural disasters in November 15, 2018 in Pohang, North Gyeongsang Province, S... | [
"\"earthquake in Pohang, South Korea, North Gyeongsang Province on 2017, exactly 15 days ago\"",
"\"earthquake in November 15, 2017 at 15:00 in Pohang, North Gyeongsang Province, South Korea\"",
"\"North Korean earthquake in November in South Korea\"",
"\"November 2017 earthquake in Pohang, South Korea\"",
... | [
"\"protesters in Pohang, South Korea, North Gyeongsang Province during 2017 Pohang earthquake\"",
"\"Protesters march in the streets of Pohang, South Korea on November 15, 2017 during a protest.\"",
"Protesters marching in the streets of Pohang, South Korea, North Gyeongsang Province, in November",
"\"Protest... | {
"article_url": "https://ko.wikipedia.org/wiki/2017%EB%85%84_%ED%8F%AC%ED%95%AD_%EC%A7%80%EC%A7%84",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe given query is to extract events that could be visualized in a video and are likely to be related to the 2017 Pohang earthquake. We should consider events that could be seen in a video of the event.\n\nGiven the nature of the query, it's likely that the event could be visualized in a video if it involves significant damage or destruction, people affected by the earthquake, or related to the event itself. The event that makes sense to extract is \"Protesters marching in the streets\" because it is a plausible and concrete event that could be seen in a video of the event. This event is related to the query, which is about the 2017 Pohang earthquake. \n\nEVENTS:\n1. Protesters marching in the streets\n2. People affected by the earthquake\n3. Damage to structures\n4. People injured\n5. Rescue efforts",
"en_article_excerpt": "The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.",
"en_article_url": "https://en.wikipedia.org/wiki/2017_Pohang_earthquake",
"event_category": "disasters",
"event_name": "pohang_earthquake",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThe 2017 Pohang earthquake is a strong seismic event that occurred in South Korea. Given the nature of the query, it is reasonable to extract events that are related to natural disasters, particularly those that are publicly documented and visually verifiable. I will consider the query as a whole and extract events that are likely to be visualized in a video, such as news reports, images, or footage of the aftermath of the earthquake.\n\nEVENTS:\n1. Aftershocks\n2. Emergency response efforts\n3. Rescue operations\n4. Destruction and damage to buildings\n5. Evacuation efforts",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\n\nConsidering the query contains specific geospatial information (Korea), locations (Heunghae and Pohang), and time (November 15, 2017), it suggests that the event might be a natural disaster or a disaster that occurred in this location. Given the location and date, it is likely that the event could be visualized in a video related to natural disasters or emergencies.\n\nEVENTS:\n\n1. Earthquake\n2. Landslide\n3. Tsunami",
"video_URL": "http://youtube.com/watch?v=6HhGFKbHFbs",
"video_description": "경북 포항에서 규모 5.4의 지진과 여진이 발생, 전국이 지진파에 흔들거리면서 주민들이 불안에 떨고 있다. 15일 기상청에 따르면 이날 오후 2시29분쯤 포항시 북구 흥해 6㎞ 지점에서 지진이 발생했다. http://www.hankookilbo.com/v/38f7f1c1597443e5ac55aca087a5dae2",
"video_id": "6HhGFKbHFbs"
} | 16 | [
"The image shows a street view captured by a security camera, displaying a white pickup truck parked on the side of a road surrounded by trees and houses. There is text in Korean that translates to \"A driver and passenger were born in a traffic accident in Gyeongbuk Poheung\" and \"The accident occurred yesterday ... | 6HhGFKbHFbs | A video captured a series of frames depicting a traffic accident in Gyeongbuk Poheung, South Korea. The accident occurred on November 15th, and the video shows a white pickup truck involved in a collision with a vehicle. The driver and passenger were injured and escaped unharmed. The video also shows the aftermath of the accident, including the police arrival and a group of pedestrians chasing the perpetrator away. The footage appears to be a news report, likely from a local news channel, as indicated by the text "포항 지진 발생 당시 영상 오늘 (11월 15일)" ("Video of the time of the Pohang earthquake today, November 15th"). The video also shows footage of the daily life of the people in the street, capturing the resilience and normalcy of the affected community. |
The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017. | [
"\"2017 aftershocks in Pohang, South Korea, North Gyeongsang Province\"",
"' aftershocks and aftershocks in November 15, 2017, in Pohang, North Gyeongsang Province, South Korea'",
"Aftershock in November in Pohang, South Korea",
"\"Natural disasters in November 15, 2018 in Pohang, North Gyeongsang Province, S... | [
"\"earthquake in Pohang, South Korea, North Gyeongsang Province on 2017, exactly 15 days ago\"",
"\"earthquake in November 15, 2017 at 15:00 in Pohang, North Gyeongsang Province, South Korea\"",
"\"North Korean earthquake in November in South Korea\"",
"\"November 2017 earthquake in Pohang, South Korea\"",
... | [
"\"protesters in Pohang, South Korea, North Gyeongsang Province during 2017 Pohang earthquake\"",
"\"Protesters march in the streets of Pohang, South Korea on November 15, 2017 during a protest.\"",
"Protesters marching in the streets of Pohang, South Korea, North Gyeongsang Province, in November",
"\"Protest... | {
"article_url": "https://ko.wikipedia.org/wiki/2017%EB%85%84_%ED%8F%AC%ED%95%AD_%EC%A7%80%EC%A7%84",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe given query is to extract events that could be visualized in a video and are likely to be related to the 2017 Pohang earthquake. We should consider events that could be seen in a video of the event.\n\nGiven the nature of the query, it's likely that the event could be visualized in a video if it involves significant damage or destruction, people affected by the earthquake, or related to the event itself. The event that makes sense to extract is \"Protesters marching in the streets\" because it is a plausible and concrete event that could be seen in a video of the event. This event is related to the query, which is about the 2017 Pohang earthquake. \n\nEVENTS:\n1. Protesters marching in the streets\n2. People affected by the earthquake\n3. Damage to structures\n4. People injured\n5. Rescue efforts",
"en_article_excerpt": "The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.",
"en_article_url": "https://en.wikipedia.org/wiki/2017_Pohang_earthquake",
"event_category": "disasters",
"event_name": "pohang_earthquake",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThe 2017 Pohang earthquake is a strong seismic event that occurred in South Korea. Given the nature of the query, it is reasonable to extract events that are related to natural disasters, particularly those that are publicly documented and visually verifiable. I will consider the query as a whole and extract events that are likely to be visualized in a video, such as news reports, images, or footage of the aftermath of the earthquake.\n\nEVENTS:\n1. Aftershocks\n2. Emergency response efforts\n3. Rescue operations\n4. Destruction and damage to buildings\n5. Evacuation efforts",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\n\nConsidering the query contains specific geospatial information (Korea), locations (Heunghae and Pohang), and time (November 15, 2017), it suggests that the event might be a natural disaster or a disaster that occurred in this location. Given the location and date, it is likely that the event could be visualized in a video related to natural disasters or emergencies.\n\nEVENTS:\n\n1. Earthquake\n2. Landslide\n3. Tsunami",
"video_URL": "http://youtube.com/watch?v=1sXMvTYhnn0",
"video_description": "+--------------------------------------------------------------------------------+ ◈ 중앙일보 공식 SNS 채널 ◈ - 페이스북 : https://www.facebook.com/joongang/ - 인스타그램 : https://www.instagram.com/joongangilbo - 트위터 : http://www.twitter.com/joongangilbo - 유튜브 : https://goo.gl/DyGoPC - 네이버 TV : http://tvcast.naver.com/joongangsisa - 네이버포스트 : https://goo.gl/eeYh4k 당신이 원하는 뉴스, 중앙일보 SNS 채널과 함께하세요. +--------------------------------------------------------------------------------+",
"video_id": "1sXMvTYhnn0"
} | 16 | [
"The image shows the exterior of a building with a glass door and windows. There is a pile of broken glass and debris on the ground near a wooden box outside the entrance. The text at the bottom of the image in Korean states that an incident occurred in Gyeongbokgung, Buk-gu, where a 5.4 magnitude earthquake caused... | 1sXMvTYhnn0 | The video describes a series of frames that depict a seismic event, specifically an earthquake, in the region of Gangneung, South Korea. The event causes damage to the building exterior and interior, with debris and broken glass scattered on the ground near the entrance. However, the interior of the building remains relatively stable, with a grandfather clock still standing despite the seismic activity.
The scene transitions through different locations, including a street scene in Pohang, a train station, and inside a transit area. The earthquake is mentioned in Korean text, but no further details are provided. The overall tone is one of caution and concern, with residents taking precautions and authorities discussing safety measures to prevent damage and ensure public safety.
The video concludes with a black screen featuring a title and credits, marking the end of the segment. |
The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017. | [
"\"2017 aftershocks in Pohang, South Korea, North Gyeongsang Province\"",
"' aftershocks and aftershocks in November 15, 2017, in Pohang, North Gyeongsang Province, South Korea'",
"Aftershock in November in Pohang, South Korea",
"\"Natural disasters in November 15, 2018 in Pohang, North Gyeongsang Province, S... | [
"\"earthquake in Pohang, South Korea, North Gyeongsang Province on 2017, exactly 15 days ago\"",
"\"earthquake in November 15, 2017 at 15:00 in Pohang, North Gyeongsang Province, South Korea\"",
"\"North Korean earthquake in November in South Korea\"",
"\"November 2017 earthquake in Pohang, South Korea\"",
... | [
"\"protesters in Pohang, South Korea, North Gyeongsang Province during 2017 Pohang earthquake\"",
"\"Protesters march in the streets of Pohang, South Korea on November 15, 2017 during a protest.\"",
"Protesters marching in the streets of Pohang, South Korea, North Gyeongsang Province, in November",
"\"Protest... | {
"article_url": "https://ko.wikipedia.org/wiki/2017%EB%85%84_%ED%8F%AC%ED%95%AD_%EC%A7%80%EC%A7%84",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe given query is to extract events that could be visualized in a video and are likely to be related to the 2017 Pohang earthquake. We should consider events that could be seen in a video of the event.\n\nGiven the nature of the query, it's likely that the event could be visualized in a video if it involves significant damage or destruction, people affected by the earthquake, or related to the event itself. The event that makes sense to extract is \"Protesters marching in the streets\" because it is a plausible and concrete event that could be seen in a video of the event. This event is related to the query, which is about the 2017 Pohang earthquake. \n\nEVENTS:\n1. Protesters marching in the streets\n2. People affected by the earthquake\n3. Damage to structures\n4. People injured\n5. Rescue efforts",
"en_article_excerpt": "The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.",
"en_article_url": "https://en.wikipedia.org/wiki/2017_Pohang_earthquake",
"event_category": "disasters",
"event_name": "pohang_earthquake",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThe 2017 Pohang earthquake is a strong seismic event that occurred in South Korea. Given the nature of the query, it is reasonable to extract events that are related to natural disasters, particularly those that are publicly documented and visually verifiable. I will consider the query as a whole and extract events that are likely to be visualized in a video, such as news reports, images, or footage of the aftermath of the earthquake.\n\nEVENTS:\n1. Aftershocks\n2. Emergency response efforts\n3. Rescue operations\n4. Destruction and damage to buildings\n5. Evacuation efforts",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\n\nConsidering the query contains specific geospatial information (Korea), locations (Heunghae and Pohang), and time (November 15, 2017), it suggests that the event might be a natural disaster or a disaster that occurred in this location. Given the location and date, it is likely that the event could be visualized in a video related to natural disasters or emergencies.\n\nEVENTS:\n\n1. Earthquake\n2. Landslide\n3. Tsunami",
"video_URL": "http://youtube.com/watch?v=MH6_TeWDitQ",
"video_description": "포항 북구 북쪽 9Km 지역서 5.4 지진 이어 여진…전국서 감지 2.2, 2.6 전진 이어 3.6 여진도 발생 기상청은 15일 오후 2시 29분 쯤 경북 포항시 북구 북쪽 9㎞ 지역에서 규모 5.4의 지진이 발생했다고 밝혔습니다. 진앙은 북위 36.10도, 동경 129.37도입니다. 기상청은 이 지진의 깊이를 9㎞로 파악했습니다. 기상청은 당초 지진 발생지역과 규모를 포항시 북구 북쪽 6㎞ 지점, 규모 5.5로 발표했으나 잠시 후 수정하는 등 혼선이 이어지고 있습니다. 이날 오후 2시49분에는 포항시 북구 북쪽 7km에서 규모 3.6의 여진도 발생했습니다. 기상청은 \"이동속도가 빠른 지진파(P파)만을 이용해 자동 추정한 정보\"라며 \"여진 등 안전에 주의해야 한다\"고 당부했습니다. 이날 지진으로 경북과 경남은 물론 서울 지역에서도 건물 흔들림이 느껴지는 등 전국 곳곳에서 진동이 감지됐습니다. 이보다 앞서 이 인근에서는 두 차례 지진이 있었다고 기상청은 발표했습니다. 오후 2시 22분 32초 포항시 북구 북쪽 7km 지역(북위 36.11도, 동경 129.36도)에서 규모 2.2의 지진이 난 후 2시 22분 44초에는 비슷한 지점(북위 36.08도, 동경 129.31도)에서 규모 2.6의 지진이 발생했습니다. (끝)",
"video_id": "MH6_TeWDitQ"
} | 16 | [
"The image is a map highlighting a specific location with a red circle. The text is in Korean, indicating various regions such as \"포항시북구\" (Pohang-si Buk-gu), \"포항시남구\" (Pohang-si Nam-gu), \"경주시\" (Gyeongju-si), and \"영천시\" (Yeongcheon-si). The red circle is centered on Pohang-si Buk-gu, suggesting a significant e... | MH6_TeWDitQ | The video primarily focuses on a gathering or assembly of people in the area of Pohang-si Buk-gu, with a strong emphasis on an event or development taking place within this region. The event seems to be centered around a news broadcast or event that involves the city of Pohang-si Buk-gu, given the presence of a news channel or network ("연암뉴스TV") and the focus on the region.
The individual gathering appears to be a planned or organized event, with a clear leader or figure standing in the center of the group. The group is organized and orderly, with a distinct line or formation. However, a significant event or disturbance occurs, resulting in the dispersal of the group and the scattering of debris across the road.
The event appears to be ongoing, with debris still visible on the road, suggesting a recent disturbance or incident. Despite the absence of military personnel, the presence of individuals in the scene and the scattered debris reinforces the idea that the event is still active. The context suggests that the event is related to the city's politics or infrastructure, but the specific details of the event or its impact on the region are not provided. |
The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017. | [
"\"2017 aftershocks in Pohang, South Korea, North Gyeongsang Province\"",
"' aftershocks and aftershocks in November 15, 2017, in Pohang, North Gyeongsang Province, South Korea'",
"Aftershock in November in Pohang, South Korea",
"\"Natural disasters in November 15, 2018 in Pohang, North Gyeongsang Province, S... | [
"\"earthquake in Pohang, South Korea, North Gyeongsang Province on 2017, exactly 15 days ago\"",
"\"earthquake in November 15, 2017 at 15:00 in Pohang, North Gyeongsang Province, South Korea\"",
"\"North Korean earthquake in November in South Korea\"",
"\"November 2017 earthquake in Pohang, South Korea\"",
... | [
"\"protesters in Pohang, South Korea, North Gyeongsang Province during 2017 Pohang earthquake\"",
"\"Protesters march in the streets of Pohang, South Korea on November 15, 2017 during a protest.\"",
"Protesters marching in the streets of Pohang, South Korea, North Gyeongsang Province, in November",
"\"Protest... | {
"article_url": "https://ko.wikipedia.org/wiki/2017%EB%85%84_%ED%8F%AC%ED%95%AD_%EC%A7%80%EC%A7%84",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe given query is to extract events that could be visualized in a video and are likely to be related to the 2017 Pohang earthquake. We should consider events that could be seen in a video of the event.\n\nGiven the nature of the query, it's likely that the event could be visualized in a video if it involves significant damage or destruction, people affected by the earthquake, or related to the event itself. The event that makes sense to extract is \"Protesters marching in the streets\" because it is a plausible and concrete event that could be seen in a video of the event. This event is related to the query, which is about the 2017 Pohang earthquake. \n\nEVENTS:\n1. Protesters marching in the streets\n2. People affected by the earthquake\n3. Damage to structures\n4. People injured\n5. Rescue efforts",
"en_article_excerpt": "The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.",
"en_article_url": "https://en.wikipedia.org/wiki/2017_Pohang_earthquake",
"event_category": "disasters",
"event_name": "pohang_earthquake",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThe 2017 Pohang earthquake is a strong seismic event that occurred in South Korea. Given the nature of the query, it is reasonable to extract events that are related to natural disasters, particularly those that are publicly documented and visually verifiable. I will consider the query as a whole and extract events that are likely to be visualized in a video, such as news reports, images, or footage of the aftermath of the earthquake.\n\nEVENTS:\n1. Aftershocks\n2. Emergency response efforts\n3. Rescue operations\n4. Destruction and damage to buildings\n5. Evacuation efforts",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\n\nConsidering the query contains specific geospatial information (Korea), locations (Heunghae and Pohang), and time (November 15, 2017), it suggests that the event might be a natural disaster or a disaster that occurred in this location. Given the location and date, it is likely that the event could be visualized in a video related to natural disasters or emergencies.\n\nEVENTS:\n\n1. Earthquake\n2. Landslide\n3. Tsunami",
"video_URL": "http://youtube.com/watch?v=_HN3DpC6m6I",
"video_description": "[현장] 포항 지진 피해 상황은…\"여진 등 안전 주의\" 오늘 오후 2시 29분 쯤 경북 포항시 북구에서 규모 5.4의 지진이 발생했습니다. 이번 지진은 작년 9월 12일 경북 경주시에서 발생한 규모 5.8의 지진에 이어 우리나라에서 일어난 지진 중 역대 두번째 규모입니다. 기상청에 따르면 포항지진 발생 지점은 포항시 북구 북쪽 9㎞, 북위 36.10도, 동경 129.37도입니다. 기상청은 이 지진의 깊이를 9㎞로 파악했습니다. 기상청은 당초 지진 발생지역과 규모를 포항시 북구 북쪽 6㎞ 지점, 규모 5.5로 발표했으나 지진 규모를 하향조정하는 등 혼선을 겪었습니다. 규모 5.4 지진에 앞서 오후 2시 22분 32초 포항시 북구 북쪽 7km 지역에서 규모 2.2, 2시 22분 44초 비슷한 지점(북위 36.08도, 동경 129.31도)에서 규모 2.6의 지진 등 전진이 발생했습니다. 아울러 이날 오후 2시 49분 규모 3.6 지진을 시작으로 오후 3시0분 54초께 규모 2.9 지진, 오후 3시 9분 49초 규모 3.6 지진 등 여진도 이어지고 있습니다. 기상청은 \"이동속도가 빠른 지진파(P파)만을 이용해 자동 추정한 정보\"라며 \"여진 등 안전에 주의해야 한다\"고 당부했습니다. 그러면서 \"계속해서 여진이 발생하고 있어 상황을 면밀히 분석 중\"이라며 \"발생 지점이 내륙 쪽으로 들어가 있어 해일의 가능성은 없는 것으로 판단한다\"고 덧붙였습니다. 이날 지진으로 경북과 경남은 물론 서울 지역에서도 건물 흔들림이 느껴지는 등 전국 곳곳에서 진동이 감지됐습니다. 연합뉴스TV : 02-398-4441(기사문의ㆍ제보) 카톡/라인 jebo23 ▣ 연합뉴스TV 유튜브 채널 구독 https://goo.gl/VuCJMi ▣ 대한민국 뉴스의 시작 연합뉴스TV / Yonhap News TV http://www.yonhapnewstv.co.kr/",
"video_id": "_HN3DpC6m6I"
} | 16 | [
"The image shows an aerial view of a densely populated urban area with numerous high-rise buildings and a large train station or transportation hub. The text in the image indicates it is from a Korean TV broadcast and mentions \"제보영상\" which translates to \"reported video.\" The specific time \"15일 오전 2시 29분\" tran... | _HN3DpC6m6I | A magnitude 5.4 earthquake occurred in Pohang City, Gyeongbuk Province, South Korea, causing widespread damage and destruction. The event was reported by a Korean TV broadcast, known as "제보 영상", and live footage showed the aftermath of the earthquake. Initially, the video depicted a cityscape and transportation infrastructure, followed by a series of images showing the earthquake's impact, including a shattered window in an Apple store and debris scattered across the floor. The footage continued to show a damaged electronics store, a crowd of people gathered in a public area, and a large crowd responding to the seismic events. The Korea Meteorological Administration's advisory was displayed on screen, emphasizing the need for caution and vigilance. Throughout the video, participants were shown engaging in various activities, including discussing the event and gathering information, indicating a collective effort to stay informed and alert. |
The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017. | [
"\"2017 aftershocks in Pohang, South Korea, North Gyeongsang Province\"",
"' aftershocks and aftershocks in November 15, 2017, in Pohang, North Gyeongsang Province, South Korea'",
"Aftershock in November in Pohang, South Korea",
"\"Natural disasters in November 15, 2018 in Pohang, North Gyeongsang Province, S... | [
"\"earthquake in Pohang, South Korea, North Gyeongsang Province on 2017, exactly 15 days ago\"",
"\"earthquake in November 15, 2017 at 15:00 in Pohang, North Gyeongsang Province, South Korea\"",
"\"North Korean earthquake in November in South Korea\"",
"\"November 2017 earthquake in Pohang, South Korea\"",
... | [
"\"protesters in Pohang, South Korea, North Gyeongsang Province during 2017 Pohang earthquake\"",
"\"Protesters march in the streets of Pohang, South Korea on November 15, 2017 during a protest.\"",
"Protesters marching in the streets of Pohang, South Korea, North Gyeongsang Province, in November",
"\"Protest... | {
"article_url": "https://ko.wikipedia.org/wiki/2017%EB%85%84_%ED%8F%AC%ED%95%AD_%EC%A7%80%EC%A7%84",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe given query is to extract events that could be visualized in a video and are likely to be related to the 2017 Pohang earthquake. We should consider events that could be seen in a video of the event.\n\nGiven the nature of the query, it's likely that the event could be visualized in a video if it involves significant damage or destruction, people affected by the earthquake, or related to the event itself. The event that makes sense to extract is \"Protesters marching in the streets\" because it is a plausible and concrete event that could be seen in a video of the event. This event is related to the query, which is about the 2017 Pohang earthquake. \n\nEVENTS:\n1. Protesters marching in the streets\n2. People affected by the earthquake\n3. Damage to structures\n4. People injured\n5. Rescue efforts",
"en_article_excerpt": "The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.",
"en_article_url": "https://en.wikipedia.org/wiki/2017_Pohang_earthquake",
"event_category": "disasters",
"event_name": "pohang_earthquake",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThe 2017 Pohang earthquake is a strong seismic event that occurred in South Korea. Given the nature of the query, it is reasonable to extract events that are related to natural disasters, particularly those that are publicly documented and visually verifiable. I will consider the query as a whole and extract events that are likely to be visualized in a video, such as news reports, images, or footage of the aftermath of the earthquake.\n\nEVENTS:\n1. Aftershocks\n2. Emergency response efforts\n3. Rescue operations\n4. Destruction and damage to buildings\n5. Evacuation efforts",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\n\nConsidering the query contains specific geospatial information (Korea), locations (Heunghae and Pohang), and time (November 15, 2017), it suggests that the event might be a natural disaster or a disaster that occurred in this location. Given the location and date, it is likely that the event could be visualized in a video related to natural disasters or emergencies.\n\nEVENTS:\n\n1. Earthquake\n2. Landslide\n3. Tsunami",
"video_URL": "http://youtube.com/watch?v=bADRfbUZewY",
"video_description": "산림청은 오는 4일 포항산림조합 목재유통센터에서 국유림에서 생산된 임산 연료를 지진피해를 입은 포항 주민들에게 무상으로 공급하는 행사를 갖기로 했습니다. ◀ＥＮＤ▶ 지진피해 주민과 홀몸노인, 기초생활보장 수급자 등을 대상으로 가구당 1톤씩 모두 21가구를 지원합니다. 이번에 공급하는 장작은 봉화군 장군봉 선도산림경영단지 내 금강소나무 육성사업지에서 생산된 참나무와 벚나무, 박달나무 등입니다.",
"video_id": "bADRfbUZewY"
} | 16 | [
"A news anchor is presenting a segment in a studio. The text on the screen indicates that the news is about Sanlim, who has been donated to a local community for support.",
"The video displays an image of a modern building, likely a public facility or community center. The building has a large banner above the en... | bADRfbUZewY | The video depicts a scene of individuals from the Forest Service providing relief support to earthquake-affected households in a community. The scene is characterized by the distribution of various materials such as logs, firewood, and construction materials, which are part of a broader relief initiative aimed at providing aid to those affected by the earthquake. The relief efforts are led by a team of individuals wearing red uniforms and helmets, who are actively engaged in clearing debris, distributing aid, and organizing the relief materials. The scene emphasizes the organized and practical nature of the aid provided by the Forest Service to support the affected communities. The relief efforts continue to focus on providing essential resources to help the households recover from the disaster. |
The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017. | [
"\"2017 aftershocks in Pohang, South Korea, North Gyeongsang Province\"",
"' aftershocks and aftershocks in November 15, 2017, in Pohang, North Gyeongsang Province, South Korea'",
"Aftershock in November in Pohang, South Korea",
"\"Natural disasters in November 15, 2018 in Pohang, North Gyeongsang Province, S... | [
"\"earthquake in Pohang, South Korea, North Gyeongsang Province on 2017, exactly 15 days ago\"",
"\"earthquake in November 15, 2017 at 15:00 in Pohang, North Gyeongsang Province, South Korea\"",
"\"North Korean earthquake in November in South Korea\"",
"\"November 2017 earthquake in Pohang, South Korea\"",
... | [
"\"protesters in Pohang, South Korea, North Gyeongsang Province during 2017 Pohang earthquake\"",
"\"Protesters march in the streets of Pohang, South Korea on November 15, 2017 during a protest.\"",
"Protesters marching in the streets of Pohang, South Korea, North Gyeongsang Province, in November",
"\"Protest... | {
"article_url": "https://ko.wikipedia.org/wiki/2017%EB%85%84_%ED%8F%AC%ED%95%AD_%EC%A7%80%EC%A7%84",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe given query is to extract events that could be visualized in a video and are likely to be related to the 2017 Pohang earthquake. We should consider events that could be seen in a video of the event.\n\nGiven the nature of the query, it's likely that the event could be visualized in a video if it involves significant damage or destruction, people affected by the earthquake, or related to the event itself. The event that makes sense to extract is \"Protesters marching in the streets\" because it is a plausible and concrete event that could be seen in a video of the event. This event is related to the query, which is about the 2017 Pohang earthquake. \n\nEVENTS:\n1. Protesters marching in the streets\n2. People affected by the earthquake\n3. Damage to structures\n4. People injured\n5. Rescue efforts",
"en_article_excerpt": "The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.",
"en_article_url": "https://en.wikipedia.org/wiki/2017_Pohang_earthquake",
"event_category": "disasters",
"event_name": "pohang_earthquake",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThe 2017 Pohang earthquake is a strong seismic event that occurred in South Korea. Given the nature of the query, it is reasonable to extract events that are related to natural disasters, particularly those that are publicly documented and visually verifiable. I will consider the query as a whole and extract events that are likely to be visualized in a video, such as news reports, images, or footage of the aftermath of the earthquake.\n\nEVENTS:\n1. Aftershocks\n2. Emergency response efforts\n3. Rescue operations\n4. Destruction and damage to buildings\n5. Evacuation efforts",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\n\nConsidering the query contains specific geospatial information (Korea), locations (Heunghae and Pohang), and time (November 15, 2017), it suggests that the event might be a natural disaster or a disaster that occurred in this location. Given the location and date, it is likely that the event could be visualized in a video related to natural disasters or emergencies.\n\nEVENTS:\n\n1. Earthquake\n2. Landslide\n3. Tsunami",
"video_URL": "http://youtube.com/watch?v=4eNANKub4NE",
"video_description": "7000미라클_포항 지진 피해 돕기 특집방송 1부 본방 2017년 11월 28일(화) 오전 10시 10분 재방 2017년 11월 28일(화) 밤 11시 10분 삼방 2017년 11월 29일(수) 오후 2시 50분 ----------------------------------------------------------------------------------------- 7000미라클_포항 지진 피해 돕기 특집방송 2부 본방 2017년 11월 29일(화) 오전 10시 10분 재방 2017년 11월 29일(화) 밤 11시 10분 삼방 2017년 11월 30일(수) 오후 2시 50분",
"video_id": "4eNANKub4NE"
} | 16 | [
"The image shows a scene of destruction, likely following a natural disaster or similar event. In the background, there is significant damage to a building, with debris scattered around and parts of the structure collapsed. The foreground features a narrow road or path, bordered by a wall and some vegetation on the... | 4eNANKub4NE | The video depicts a series of images capturing the aftermath of a natural disaster, particularly an earthquake. The context is that of a devastating earthquake that resulted in significant damage to a building and surrounding area. The video focuses on the emotional and physical impact of the disaster on residents and aid workers, including their efforts to provide relief and support.
Key events and participants included:
- A group of aid workers organizing supplies and distributing them in an outdoor setting.
- A church community coming together for a fundraising event to support relief efforts.
- The personal reflections of affected individuals sharing their experiences and emotions.
- The church as a source of comfort and guidance for the affected community.
Overall, the video highlights the human impact of disaster and the importance of community support and resilience in the face of adversity. |
The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017. | [
"\"2017 aftershocks in Pohang, South Korea, North Gyeongsang Province\"",
"' aftershocks and aftershocks in November 15, 2017, in Pohang, North Gyeongsang Province, South Korea'",
"Aftershock in November in Pohang, South Korea",
"\"Natural disasters in November 15, 2018 in Pohang, North Gyeongsang Province, S... | [
"\"earthquake in Pohang, South Korea, North Gyeongsang Province on 2017, exactly 15 days ago\"",
"\"earthquake in November 15, 2017 at 15:00 in Pohang, North Gyeongsang Province, South Korea\"",
"\"North Korean earthquake in November in South Korea\"",
"\"November 2017 earthquake in Pohang, South Korea\"",
... | [
"\"protesters in Pohang, South Korea, North Gyeongsang Province during 2017 Pohang earthquake\"",
"\"Protesters march in the streets of Pohang, South Korea on November 15, 2017 during a protest.\"",
"Protesters marching in the streets of Pohang, South Korea, North Gyeongsang Province, in November",
"\"Protest... | {
"article_url": "https://ko.wikipedia.org/wiki/2017%EB%85%84_%ED%8F%AC%ED%95%AD_%EC%A7%80%EC%A7%84",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe given query is to extract events that could be visualized in a video and are likely to be related to the 2017 Pohang earthquake. We should consider events that could be seen in a video of the event.\n\nGiven the nature of the query, it's likely that the event could be visualized in a video if it involves significant damage or destruction, people affected by the earthquake, or related to the event itself. The event that makes sense to extract is \"Protesters marching in the streets\" because it is a plausible and concrete event that could be seen in a video of the event. This event is related to the query, which is about the 2017 Pohang earthquake. \n\nEVENTS:\n1. Protesters marching in the streets\n2. People affected by the earthquake\n3. Damage to structures\n4. People injured\n5. Rescue efforts",
"en_article_excerpt": "The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.",
"en_article_url": "https://en.wikipedia.org/wiki/2017_Pohang_earthquake",
"event_category": "disasters",
"event_name": "pohang_earthquake",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThe 2017 Pohang earthquake is a strong seismic event that occurred in South Korea. Given the nature of the query, it is reasonable to extract events that are related to natural disasters, particularly those that are publicly documented and visually verifiable. I will consider the query as a whole and extract events that are likely to be visualized in a video, such as news reports, images, or footage of the aftermath of the earthquake.\n\nEVENTS:\n1. Aftershocks\n2. Emergency response efforts\n3. Rescue operations\n4. Destruction and damage to buildings\n5. Evacuation efforts",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\n\nConsidering the query contains specific geospatial information (Korea), locations (Heunghae and Pohang), and time (November 15, 2017), it suggests that the event might be a natural disaster or a disaster that occurred in this location. Given the location and date, it is likely that the event could be visualized in a video related to natural disasters or emergencies.\n\nEVENTS:\n\n1. Earthquake\n2. Landslide\n3. Tsunami",
"video_URL": "http://youtube.com/watch?v=QIkpgLcEWMA",
"video_description": "픽업영상|시민들이 보내 준 #포항지진_영상 한동대 외벽 무너지고, 주차장 바닥 갈라져 #JTBC는_여러분들의_제보_기다립니다 ---------------------------------------------------------------------------------------- 📌 소셜스토리 - JTBC 바로가기▼ https://www.facebook.com/JTBCstandbyyou/",
"video_id": "QIkpgLcEWMA"
} | 16 | [
"The video frame shows a view through trees with the text \"소셜스토리\" in the top left and \"JTBC 시청자제보\" in the top right. The main text in the center reads \"졸아진 벽돌 위험한 상황,\" which translates to \"Dangerous situation on a collapsed brick wall.\" The scene appears to depict a potentially hazardous area with trees in ... | QIkpgLcEWMA | The video depicts a series of events and scenes related to a story about students taking refuge against a collapsing building facade, with a CCTV camera capturing the situation. The story unfolds through a series of frames showing the scene from multiple angles, including inside a hair salon, a restaurant, and a parking lot. The events and interactions are described in detail, with scenes showing the students, staff, and possibly viewers reacting to the situation. The footage also includes the text "소셜 스토리" (Social Story) and "시청자제보" (Viewer Report), indicating that the clip was submitted by viewers. Throughout the sequence, the setting remains a building facade, with the camera capturing the aftermath of a disaster, while the text "현재도 여진 계속 이어져 주의 필요" (There is a need for caution as aftershocks continue) suggests that the event is ongoing. |
The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017. | [
"\"2017 aftershocks in Pohang, South Korea, North Gyeongsang Province\"",
"' aftershocks and aftershocks in November 15, 2017, in Pohang, North Gyeongsang Province, South Korea'",
"Aftershock in November in Pohang, South Korea",
"\"Natural disasters in November 15, 2018 in Pohang, North Gyeongsang Province, S... | [
"\"earthquake in Pohang, South Korea, North Gyeongsang Province on 2017, exactly 15 days ago\"",
"\"earthquake in November 15, 2017 at 15:00 in Pohang, North Gyeongsang Province, South Korea\"",
"\"North Korean earthquake in November in South Korea\"",
"\"November 2017 earthquake in Pohang, South Korea\"",
... | [
"\"protesters in Pohang, South Korea, North Gyeongsang Province during 2017 Pohang earthquake\"",
"\"Protesters march in the streets of Pohang, South Korea on November 15, 2017 during a protest.\"",
"Protesters marching in the streets of Pohang, South Korea, North Gyeongsang Province, in November",
"\"Protest... | {
"article_url": "https://ko.wikipedia.org/wiki/2017%EB%85%84_%ED%8F%AC%ED%95%AD_%EC%A7%80%EC%A7%84",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe given query is to extract events that could be visualized in a video and are likely to be related to the 2017 Pohang earthquake. We should consider events that could be seen in a video of the event.\n\nGiven the nature of the query, it's likely that the event could be visualized in a video if it involves significant damage or destruction, people affected by the earthquake, or related to the event itself. The event that makes sense to extract is \"Protesters marching in the streets\" because it is a plausible and concrete event that could be seen in a video of the event. This event is related to the query, which is about the 2017 Pohang earthquake. \n\nEVENTS:\n1. Protesters marching in the streets\n2. People affected by the earthquake\n3. Damage to structures\n4. People injured\n5. Rescue efforts",
"en_article_excerpt": "The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.",
"en_article_url": "https://en.wikipedia.org/wiki/2017_Pohang_earthquake",
"event_category": "disasters",
"event_name": "pohang_earthquake",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThe 2017 Pohang earthquake is a strong seismic event that occurred in South Korea. Given the nature of the query, it is reasonable to extract events that are related to natural disasters, particularly those that are publicly documented and visually verifiable. I will consider the query as a whole and extract events that are likely to be visualized in a video, such as news reports, images, or footage of the aftermath of the earthquake.\n\nEVENTS:\n1. Aftershocks\n2. Emergency response efforts\n3. Rescue operations\n4. Destruction and damage to buildings\n5. Evacuation efforts",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\n\nConsidering the query contains specific geospatial information (Korea), locations (Heunghae and Pohang), and time (November 15, 2017), it suggests that the event might be a natural disaster or a disaster that occurred in this location. Given the location and date, it is likely that the event could be visualized in a video related to natural disasters or emergencies.\n\nEVENTS:\n\n1. Earthquake\n2. Landslide\n3. Tsunami",
"video_URL": "http://youtube.com/watch?v=_vMNNh-x79Q",
"video_description": "경북 포항서 규모 5.4 지진 발생…서울ㆍ부산서도 진동 [앵커] 오늘 오후 경북 포항에서 규모 5.4의 지진이 발생했습니다. 부산을 비롯해 서울에서도 흔들림이 감지됐습니다. 보도국 취재기자 연결해 자세한 내용 들어보겠습니다. 홍정원 기자. [기자] 네, 오늘 오후 2시29분 포항 북구에서 북쪽으로 6킬로미터 지역에서 규모 5.4의 지진이 발생했다고 기상청이 밝혔습니다. 진앙은 북위 36.10도, 동경 129.37도입니다. 포항에서는 창이 흔들리고 책장에서 책이 떨어지는 등 진동의 여파가 상당했던 것으로 전해지고 있는데요. 또한 포항 곳곳에선 여진도 이어진 것으로 전해지고 있습니다. 기상청은 \"이동속도가 빠른 지진파, P파만을 이용해 자동 추정한 정보\"라며 \"여진 등 안전에 주의해야 한다\"고 당부했습니다. 한수원 측은 전국의 원자력발전소는 별다른 문제 없이 가동 중이라고 밝혔습니다. 오늘 지진의 지진동은 서울 등 경기 남부는 물론, 경남권과 호남권 등 전국적으로 감지된 것으로 전해지고 있습니다. 지진 발생 직후 부산소방안전본부와 부산경찰청에는 건물이 흔들린다는 시민들의 문의 전화가 빗발친 것으로 전해졌습니다. 일부 도심 건물이 심하게 흔들리면서 관련 직장인과 시민들이 급히 대피하는 소동이 벌어졌습니다. 다시 한 번 말씀드리겠습니다. 오늘 오후 2시29분 포항 북구에서 북쪽으로 6킬로미터 지역에서 규모 5.4의 지진이 발생했다고 기상청이 밝혔습니다. 포항 곳곳에선 여진이 이어지면서 주민 대피가 속출했으며, 지진동은 서울 등 경기권과 부산 등에서도 감지되면서 전국적으로 우려를 키우고 있습니다. 원자력발전소를 관리하는 한수원 측은 현재까지 가동중인 원전에 별다른 문제가 없다고 밝혔습니다. 아직까지 전국적으로 정확한 인명, 시설 피해 상황은 집계되지 않은 상태입니다. 지금까지 보도국에서 전해드렸습니다. 연합뉴스TV : 02-398-4441(기사문의·제보) 카톡/라인 jebo23 (끝)",
"video_id": "_vMNNh-x79Q"
} | 16 | [
"The image shows a news broadcast with two anchors, a man and a woman, seated at a news desk. The background features a cityscape. The main headline in Korean text reads: \"교육부 '내일 수능 평균 등 전국서 예정대로 진행'.\" The broadcast is from Newstalk News TV, with a timestamp of 14:58. Additional news ticker texts are visible at ... | _vMNNh-x79Q | The video depicts a news broadcast on Newstalk News TV discussing the Ministry of Education's decision to proceed with the national college entrance exam as planned despite disruptions caused by an earthquake. The broadcast involves several anchors, a man on the phone providing commentary, and a close-up of a chandelier. The camera pans over cityscapes and buildings, showing reports of damage and debris, while the anchor desk setup remains consistent throughout. The broadcast continues to emphasize the importance of the exam and the Ministry's decision. |
The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017. | [
"\"2017 aftershocks in Pohang, South Korea, North Gyeongsang Province\"",
"' aftershocks and aftershocks in November 15, 2017, in Pohang, North Gyeongsang Province, South Korea'",
"Aftershock in November in Pohang, South Korea",
"\"Natural disasters in November 15, 2018 in Pohang, North Gyeongsang Province, S... | [
"\"earthquake in Pohang, South Korea, North Gyeongsang Province on 2017, exactly 15 days ago\"",
"\"earthquake in November 15, 2017 at 15:00 in Pohang, North Gyeongsang Province, South Korea\"",
"\"North Korean earthquake in November in South Korea\"",
"\"November 2017 earthquake in Pohang, South Korea\"",
... | [
"\"protesters in Pohang, South Korea, North Gyeongsang Province during 2017 Pohang earthquake\"",
"\"Protesters march in the streets of Pohang, South Korea on November 15, 2017 during a protest.\"",
"Protesters marching in the streets of Pohang, South Korea, North Gyeongsang Province, in November",
"\"Protest... | {
"article_url": "https://ko.wikipedia.org/wiki/2017%EB%85%84_%ED%8F%AC%ED%95%AD_%EC%A7%80%EC%A7%84",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe given query is to extract events that could be visualized in a video and are likely to be related to the 2017 Pohang earthquake. We should consider events that could be seen in a video of the event.\n\nGiven the nature of the query, it's likely that the event could be visualized in a video if it involves significant damage or destruction, people affected by the earthquake, or related to the event itself. The event that makes sense to extract is \"Protesters marching in the streets\" because it is a plausible and concrete event that could be seen in a video of the event. This event is related to the query, which is about the 2017 Pohang earthquake. \n\nEVENTS:\n1. Protesters marching in the streets\n2. People affected by the earthquake\n3. Damage to structures\n4. People injured\n5. Rescue efforts",
"en_article_excerpt": "The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.",
"en_article_url": "https://en.wikipedia.org/wiki/2017_Pohang_earthquake",
"event_category": "disasters",
"event_name": "pohang_earthquake",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThe 2017 Pohang earthquake is a strong seismic event that occurred in South Korea. Given the nature of the query, it is reasonable to extract events that are related to natural disasters, particularly those that are publicly documented and visually verifiable. I will consider the query as a whole and extract events that are likely to be visualized in a video, such as news reports, images, or footage of the aftermath of the earthquake.\n\nEVENTS:\n1. Aftershocks\n2. Emergency response efforts\n3. Rescue operations\n4. Destruction and damage to buildings\n5. Evacuation efforts",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2017 Pohang earthquake, measuring magnitude 5.4 on the Richter magnitude scale, struck Heunghae, Pohang, North Gyeongsang Province, South Korea on November 15, 2017.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\n\nConsidering the query contains specific geospatial information (Korea), locations (Heunghae and Pohang), and time (November 15, 2017), it suggests that the event might be a natural disaster or a disaster that occurred in this location. Given the location and date, it is likely that the event could be visualized in a video related to natural disasters or emergencies.\n\nEVENTS:\n\n1. Earthquake\n2. Landslide\n3. Tsunami",
"video_URL": "http://youtube.com/watch?v=k_iUdIJKUUc",
"video_description": "포항 시민들은 눈앞에서 아파트 외벽이 갈라지는 현장을 목격하며 하루종일 공포에 떨었습니다. 시민 700여명은 지금도 집에 돌아가지 못하고 대피소에 머물고 있습니다. [Ch.19] 사실을 보고 진실을 말합니다. ★홈페이지 : http://news.tvchosun.com/ * 뉴스제보 : 이메일(tvchosun@chosun.com), 카카오톡(tv조선제보), 전화(1661-0190)",
"video_id": "k_iUdIJKUUc"
} | 16 | [
"The image shows a news anchor in a suit standing in front of a blurred background that appears to depict an outdoor setting with people walking and vehicles. The news anchor is speaking, and there is Korean text on the screen that translates to \"Live TV News Channel.\" Below, another line of text in Korean reads,... | k_iUdIJKUUc | A 16-frame video depicts the aftermath of a severe natural disaster, specifically an earthquake, in Pohang, South Korea. The scene begins with a news anchor speaking on live TV, announcing an abnormal situation and stating that it's "abnormal" that it's happening. The anchor is shown standing in front of a blurred outdoor background with people walking and vehicles, indicating the chaos and destruction caused by the disaster. The news anchor speaks to two reporters, who are holding microphones and standing in a crowded indoor setting, with a gymnasium in the background.
As the disaster unfolds, the camera shows people seeking shelter, including a reporter in a red jacket holding a microphone, and a group of people carrying backpacks and walking towards a van. The scene shifts to a damaged building with a cracked exterior wall, indicating the severity of the earthquake. A person is seen walking at night, holding a bag and a book, and a large indoor space filled with displaced persons is shown, including people sitting and standing on the floor.
The video progresses to depict the aftermath, showing a large indoor space filled with people, some lying on the floor, resting or possibly sleeping, suggesting they have been sheltered for an extended period. The overall context is one of uncertainty and chaos, with the number "9" and "TV조선 LIVE" appearing throughout the frames. The final frame shows a group of people standing together, seeking help or evacuation, emphasizing the resilience and collective nature of the disaster-stricken community. |
Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan. | [] | [] | [] | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%83%9C%ED%92%8D_%EB%82%9C%EB%A7%88%EB%8F%8C_(2022%EB%85%84)",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe query is focused on a specific storm event, Typhoon Nanmadol, which is a strong typhoon. Therefore, it is reasonable to extract events related to the storm, such as the storm's name and its impact. Additionally, the query's mention of the Philippines adds specificity to the storm's location, which should help to narrow down the search results to more relevant events.\n\nEVENTS:\n1. Typhoon Nanmadol\n2. Typhoon Josie\n3. Japan\n4. Philippines\n5. Typhoon",
"en_article_excerpt": "Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.",
"en_article_url": "https://en.wikipedia.org/wiki/Typhoon_Nanmadol_(2022)",
"event_category": "disasters",
"event_name": "typhoon_nanmadol",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nBased on my analysis, I consider the following events to be relevant to the given query: Typhoon Nanmadol, Super Typhoon Josie, and related typhoons, as these events are directly related to the query, especially considering its geographical location (Philippines). I also consider the fact that Typhoon Nanmadol was a significant storm, making it a suitable event to extract.\n\nEVENTS:\n1. Typhoon Nanmadol\n2. Typhoon Josie\n3. Super Typhoon Josie\n4. Typhoon Nanmadol affects Japan",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nGiven the query \"Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan\", the relevant events that could be extracted in a video are related to the impact and damage caused by the storm. Specifically, the events should focus on the aftermath and consequences of the storm, including destruction, flooding, landslides, and loss of life.\n\nEVENTS:\n1. Typhoon Nanmadol causing widespread destruction in Japan, with landslides and floods reported\n2. Rescue efforts underway to locate survivors\n3. Emergency declarations issued by the Japanese government to evacuate people from affected areas\n4. Floodwaters and landslides affecting the city of Fukuoka\n5. Rescue workers searching for survivors in the affected areas",
"video_URL": "http://youtube.com/watch?v=Be0D3w0LKSg",
"video_description": "(서울=뉴스1) 문동주 기자 = 지난 18일 한국보다 먼저 일본에 상륙한 제14호 태풍 난마돌. 난마돌로 인해 일본에는 엄청난 강풍과 호우 피해가 이어졌다. 사람도 서 있기 힘든 정도의 위력의 비바람에 건물 외장재들이 날아다니고 대중교통 일부도 마비됐다. 동해에 접한 일본 서부 연안을 따라 이동한 난마돌의 위력을 영상으로 확인해보자. #난마돌 #태풍 #일본",
"video_id": "Be0D3w0LKSg"
} | 16 | [
"In the image, a person is standing inside a train compartment, possibly engaged in an activity. The text on the image reads \"일본의 모습?\" which translates to \"Appearance of Japan?\" indicating a focus on Japanese culture or context.",
"The current frame shows the interior of a train compartment with a muddy floor... | Be0D3w0LKSg | A severe typhoon, Typhoon Nanmadol, caused significant damage and flooding in Japan in 2018, affecting several regions. The typhoon resulted in powerful storms, flooding, and wind damage to buildings and structures. In this 16-frame video, the footage depicts the various stages of the typhoon's impact, from its formation to its aftermath. The video starts with a person standing in a train compartment, then shifts to show a muddy train floor, critiquing Japan's maintenance standards. The sequence progresses to a dramatic storm with a massive typhoon, Typhoon Nanmadol, affecting the Nama River in Korea. The video continues with severe flooding, including a bus being submerged, with passengers clinging to seats, and then a residential area flooded. The impact of the typhoon is highlighted through the presence of trees being blown by strong winds, trees being submerged, and debris-filled streets. The video also shows the impact on transportation, including flooded roads, with a vehicle driving through the water and a bus submerged with people clinging to seats. The video concludes with the news channel "뉴스1" continuing to broadcast news coverage of the typhoon's impact on the region. |
Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan. | [] | [] | [] | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%83%9C%ED%92%8D_%EB%82%9C%EB%A7%88%EB%8F%8C_(2022%EB%85%84)",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe query is focused on a specific storm event, Typhoon Nanmadol, which is a strong typhoon. Therefore, it is reasonable to extract events related to the storm, such as the storm's name and its impact. Additionally, the query's mention of the Philippines adds specificity to the storm's location, which should help to narrow down the search results to more relevant events.\n\nEVENTS:\n1. Typhoon Nanmadol\n2. Typhoon Josie\n3. Japan\n4. Philippines\n5. Typhoon",
"en_article_excerpt": "Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.",
"en_article_url": "https://en.wikipedia.org/wiki/Typhoon_Nanmadol_(2022)",
"event_category": "disasters",
"event_name": "typhoon_nanmadol",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nBased on my analysis, I consider the following events to be relevant to the given query: Typhoon Nanmadol, Super Typhoon Josie, and related typhoons, as these events are directly related to the query, especially considering its geographical location (Philippines). I also consider the fact that Typhoon Nanmadol was a significant storm, making it a suitable event to extract.\n\nEVENTS:\n1. Typhoon Nanmadol\n2. Typhoon Josie\n3. Super Typhoon Josie\n4. Typhoon Nanmadol affects Japan",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nGiven the query \"Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan\", the relevant events that could be extracted in a video are related to the impact and damage caused by the storm. Specifically, the events should focus on the aftermath and consequences of the storm, including destruction, flooding, landslides, and loss of life.\n\nEVENTS:\n1. Typhoon Nanmadol causing widespread destruction in Japan, with landslides and floods reported\n2. Rescue efforts underway to locate survivors\n3. Emergency declarations issued by the Japanese government to evacuate people from affected areas\n4. Floodwaters and landslides affecting the city of Fukuoka\n5. Rescue workers searching for survivors in the affected areas",
"video_URL": "http://youtube.com/watch?v=AXoJU3hRho0",
"video_description": "월요일인 오늘(19일)은 태풍 난마돌 의 영향으로 경상권과 강원영동 지역에 강한 비바람이 불겠습니다. 태풍14호 난마돌은 오늘(19일) 오전 3시 현재 일본 가고시마 북북서쪽 약 140km 해상에서 시속 17km로 북진할 것으로 예측되는데요.",
"video_id": "AXoJU3hRho0"
} | 16 | [
"A person is standing in front of a weather map, presenting information about a typhoon. The text on the image reads \"태풍 14호 남마돌 오늘 오후 영향권 벗어나 '경로 변경?'\" indicating that Typhoon 14, Namtheun, is expected to leave the influence area this afternoon, with a question about possible route changes.",
"A weather map di... | AXoJU3hRho0 | Typhoon 14, Namtheun, is expected to move northeast at 17 km per hour over Hokkaido and surrounding areas. The storm has transitioned from a 'strong' state to 'strong' as indicated by the increased wind speed, which is likely causing high winds and potentially dangerous weather conditions. The storm is predicted to leave an area of 240 km away from Yongheung-do by 3 PM today. |
Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan. | [] | [] | [] | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%83%9C%ED%92%8D_%EB%82%9C%EB%A7%88%EB%8F%8C_(2022%EB%85%84)",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe query is focused on a specific storm event, Typhoon Nanmadol, which is a strong typhoon. Therefore, it is reasonable to extract events related to the storm, such as the storm's name and its impact. Additionally, the query's mention of the Philippines adds specificity to the storm's location, which should help to narrow down the search results to more relevant events.\n\nEVENTS:\n1. Typhoon Nanmadol\n2. Typhoon Josie\n3. Japan\n4. Philippines\n5. Typhoon",
"en_article_excerpt": "Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.",
"en_article_url": "https://en.wikipedia.org/wiki/Typhoon_Nanmadol_(2022)",
"event_category": "disasters",
"event_name": "typhoon_nanmadol",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nBased on my analysis, I consider the following events to be relevant to the given query: Typhoon Nanmadol, Super Typhoon Josie, and related typhoons, as these events are directly related to the query, especially considering its geographical location (Philippines). I also consider the fact that Typhoon Nanmadol was a significant storm, making it a suitable event to extract.\n\nEVENTS:\n1. Typhoon Nanmadol\n2. Typhoon Josie\n3. Super Typhoon Josie\n4. Typhoon Nanmadol affects Japan",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nGiven the query \"Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan\", the relevant events that could be extracted in a video are related to the impact and damage caused by the storm. Specifically, the events should focus on the aftermath and consequences of the storm, including destruction, flooding, landslides, and loss of life.\n\nEVENTS:\n1. Typhoon Nanmadol causing widespread destruction in Japan, with landslides and floods reported\n2. Rescue efforts underway to locate survivors\n3. Emergency declarations issued by the Japanese government to evacuate people from affected areas\n4. Floodwaters and landslides affecting the city of Fukuoka\n5. Rescue workers searching for survivors in the affected areas",
"video_URL": "http://youtube.com/watch?v=MYkG8SmCWNo",
"video_description": "국제우주정거장에서 포착된 제14호 태풍 '난마돌'의 모습이 공개됐습니다. 소용돌이치는 거대한 구름 덩어리 한가운데로 뚜렷한 태풍의 눈이 보입니다. 미국 항공우주국, 나사 소속 우주 비행사인 밥 하인스가 18일, 트위터에 공개한 사진들인데요. 하인스는 국제우주정거장에서 찍은 난마돌의 사진 3장을 올리며, \"태풍 난마돌 경로에 있는 모든 사람들의 안전을 위해 기도한다\"고 적었습니다. https://imnews.imbc.com/replay/2022/nwtoday/article/6408956_35752.html #태풍, #난마돌, #국제우주정거장",
"video_id": "MYkG8SmCWNo"
} | 16 | [
"The image shows a news broadcast from MBC. The left side displays a map with a highlighted area in the ocean, indicating a location. The text in Korean mentions \"태풍 '난마돌' 일본 상륙 오전 열차 운행 조정\" which translates to \"Typhoon 'Nanmadol' lands in Japan, morning train operation adjustment.\" The right side features an a... | MYkG8SmCWNo | The video is about Typhoon Nanmadol, a powerful typhoon that landed in Japan. The video starts with a news broadcast announcing the typhoon's arrival, showing a map highlighting the location. The anchor mentions the typhoon's impact on the country and the International Space Station. The video then shifts to a series of frames showing the typhoon from various perspectives, including from space, the ground, and a news broadcast.
The frames depict the typhoon's appearance as a swirling cloud system, with images of the eye of the storm and its intense winds. The video also shows updates on the typhoon's movement and the impact on the surrounding area, including a golf tournament and weather conditions. |
Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan. | [] | [] | [] | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%83%9C%ED%92%8D_%EB%82%9C%EB%A7%88%EB%8F%8C_(2022%EB%85%84)",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe query is focused on a specific storm event, Typhoon Nanmadol, which is a strong typhoon. Therefore, it is reasonable to extract events related to the storm, such as the storm's name and its impact. Additionally, the query's mention of the Philippines adds specificity to the storm's location, which should help to narrow down the search results to more relevant events.\n\nEVENTS:\n1. Typhoon Nanmadol\n2. Typhoon Josie\n3. Japan\n4. Philippines\n5. Typhoon",
"en_article_excerpt": "Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.",
"en_article_url": "https://en.wikipedia.org/wiki/Typhoon_Nanmadol_(2022)",
"event_category": "disasters",
"event_name": "typhoon_nanmadol",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nBased on my analysis, I consider the following events to be relevant to the given query: Typhoon Nanmadol, Super Typhoon Josie, and related typhoons, as these events are directly related to the query, especially considering its geographical location (Philippines). I also consider the fact that Typhoon Nanmadol was a significant storm, making it a suitable event to extract.\n\nEVENTS:\n1. Typhoon Nanmadol\n2. Typhoon Josie\n3. Super Typhoon Josie\n4. Typhoon Nanmadol affects Japan",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nGiven the query \"Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan\", the relevant events that could be extracted in a video are related to the impact and damage caused by the storm. Specifically, the events should focus on the aftermath and consequences of the storm, including destruction, flooding, landslides, and loss of life.\n\nEVENTS:\n1. Typhoon Nanmadol causing widespread destruction in Japan, with landslides and floods reported\n2. Rescue efforts underway to locate survivors\n3. Emergency declarations issued by the Japanese government to evacuate people from affected areas\n4. Floodwaters and landslides affecting the city of Fukuoka\n5. Rescue workers searching for survivors in the affected areas",
"video_URL": "http://youtube.com/watch?v=waVeQcHP3f0",
"video_description": "\"우리나라에 미치는 영향은 그대로\"…강한 비바람 태풍 ’난마돌’, 매우 강한 위력 유지한 채 북상 비보다는 바람 강한 태풍…영남 해안에 집중호우 제주도와 영남 해안, 울릉도·독도에 35㎧ 돌풍 [앵커] 북상 중인 14호 태풍 '난마돌'의 진로가 동쪽으로 조금 더 치우치는 것으로 분석됐습니다. 하지만 태풍의 세력이 워낙 강해 제주도와 해안가에는 여전히 비바람 피해가 우려됩니다. 취재 기자 연결해 자세한 태풍 상황 알아보겠습니다. 김진두 기자! 태풍이 일본 쪽으로 더 치우쳐 이동한다는 건데, 그래도 강한 비바람은 그대로 유지된다는 거죠? [기자] 네, 그렇습니다. 기상청은 오늘 오전 브리핑에서 태풍 진로가 미세하게 조정됐다고 밝혔습니다. 태풍이 내일 새벽쯤 일본 규슈 북부에 상륙하는 건 그대로지만, 진로가 약간 더 동쪽으로 움직인다는 겁니다. 우리나라로부터 거리가 더 멀어지는 것이지만, 비바람 예보는 그대로 유지됐습니다. 태풍이 '매우 강한' 위력을 지닌 채 북상하기 때문인데요, 우선 바람이 무척 강하게 불겠고 영남 해안과 영동에는 집중호우도 예상되고 있습니다. 기상청은 제주도와 영남 해안, 그리고 울릉도·독도에 순간 최대 풍속이 35m에 달하는 돌풍이 불겠다고 밝혔습니다. 그 밖의 해안가와 경남 내륙에도 최고 초속 25m, 내륙 지방에도 초속 15m의 순간적인 강풍이 예상됩니다. 비는 영남 해안과 영동지방, 그리고 울릉도, 독도에 50~100mm가 오겠고, 특히 영남 해안에는 150mm가 넘는 호우가 쏟아질 것으로 보입니다. 기상청은 이번 태풍의 고비가 오늘 밤부터 내일 오전 사이가 되겠고, 제주도와 해안가에는 폭풍해일이나 월파 가능성도 있다며 철저한 대비를 당부했습니다. 지금까지 YTN 김진두입니다. YTN 김진두 (jdkim@ytn.co.kr) ※ '당신의 제보가 뉴스가 됩니다' [카카오톡] YTN 검색해 채널 추가 [전화] 02-398-8585 [메일] social@ytn.co.kr ▶ 기사 원문 : https://www.ytn.co.kr/_ln/0108_202209181259028146 ▶ 제보 하기 : https://mj.ytn.co.kr/mj/mj_write.php ▣ YTN 유튜브 채널 구독 : http://goo.gl/Ytb5SZ ⓒ YTN 무단 전재 및 재배포금지",
"video_id": "waVeQcHP3f0"
} | 16 | [
"The image shows two news anchors sitting at a desk, likely presenting a weather update. They are positioned in front of a screen displaying satellite imagery of a typhoon. The text on the screen translates to \"Typhoon 'Nanmadol' moves eastward... 'We will closely monitor.'\"",
"The image shows the same two news... | waVeQcHP3f0 | Typhoon 'Nanmadol' is a powerful typhoon that has been moving eastward from the East China Sea, causing significant weather conditions across various regions, including Korea and Japan. The video shows ongoing updates on the typhoon's trajectory and impact, including its projected path, wind direction, and rainfall intensity. The typhoon has caused severe weather conditions, including strong winds, heavy rainfall, and coastal flooding, with strong gale winds expected in the southern coast of Korea. The video highlights the impact of the typhoon on various aspects, such as transportation, maritime activities, and coastal infrastructure. |
Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan. | [] | [] | [] | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%83%9C%ED%92%8D_%EB%82%9C%EB%A7%88%EB%8F%8C_(2022%EB%85%84)",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe query is focused on a specific storm event, Typhoon Nanmadol, which is a strong typhoon. Therefore, it is reasonable to extract events related to the storm, such as the storm's name and its impact. Additionally, the query's mention of the Philippines adds specificity to the storm's location, which should help to narrow down the search results to more relevant events.\n\nEVENTS:\n1. Typhoon Nanmadol\n2. Typhoon Josie\n3. Japan\n4. Philippines\n5. Typhoon",
"en_article_excerpt": "Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.",
"en_article_url": "https://en.wikipedia.org/wiki/Typhoon_Nanmadol_(2022)",
"event_category": "disasters",
"event_name": "typhoon_nanmadol",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nBased on my analysis, I consider the following events to be relevant to the given query: Typhoon Nanmadol, Super Typhoon Josie, and related typhoons, as these events are directly related to the query, especially considering its geographical location (Philippines). I also consider the fact that Typhoon Nanmadol was a significant storm, making it a suitable event to extract.\n\nEVENTS:\n1. Typhoon Nanmadol\n2. Typhoon Josie\n3. Super Typhoon Josie\n4. Typhoon Nanmadol affects Japan",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nGiven the query \"Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan\", the relevant events that could be extracted in a video are related to the impact and damage caused by the storm. Specifically, the events should focus on the aftermath and consequences of the storm, including destruction, flooding, landslides, and loss of life.\n\nEVENTS:\n1. Typhoon Nanmadol causing widespread destruction in Japan, with landslides and floods reported\n2. Rescue efforts underway to locate survivors\n3. Emergency declarations issued by the Japanese government to evacuate people from affected areas\n4. Floodwaters and landslides affecting the city of Fukuoka\n5. Rescue workers searching for survivors in the affected areas",
"video_URL": "http://youtube.com/watch?v=umbAs-igFhc",
"video_description": "역대급 세력을 지닌 14호 태풍 '난마돌'이 일본에 상륙하면서 우리나라 남해와 동해 상은 물론 영남 해안에도 태풍 경보가 내려졌습니다. 현재 부산과 울산 등 태풍 특보 지역에는 초속 30m 안팎의 돌풍이 불고 있어 강풍 피해가 우려됩니다. 또 포항과 경주에는 산사태 주의보가 발령 중 입니다. 기상청은 14호 태풍 '난마돌'이 어제 저녁 일본 가고시마 남쪽에 상륙한 뒤 태풍의 세력이 매우 강에서 강한 강도로 한 단계 약화했다고 밝혔습니다. 하지만 중심 기압은 여전히 955 헥토파스칼로 강한 세력을 유지하고 있고, 중심 부근에는 초속 40m 돌풍이 몰아치고 있습니다. 특히 강풍 반경이 400km로 넓어 우리나라 남해와 동해상은 물론, 부산과 울산, 포항, 경주 등 영남 해안에도 태풍경보가 발령 중입니다. 기상청은 오늘 영남 해안에는 초속 25에서 35m의 돌풍이 불고, 영남 해안과 영동, 울릉도 독도에는 최고 100mm의 호우가 예상된다고 밝혔습니다. 특히 해상에는 최고 10m의 높은 파도가 일겠고, 해안가에는 월파나 폭풍해일 가능성이 크다며 철저한 대비를 당부했습니다. YTN 정혜윤 (jh0302@ytn.co.kr) ※ '당신의 제보가 뉴스가 됩니다' [카카오톡] YTN 검색해 채널 추가 [전화] 02-398-8585 [메일] social@ytn.co.kr ▶ 기사 원문 : https://www.ytn.co.kr/_ln/0108_202209190411440551 ▶ 제보 하기 : https://mj.ytn.co.kr/mj/mj_write.php ▣ YTN 유튜브 채널 구독 : http://goo.gl/Ytb5SZ ⓒ YTN 무단 전재 및 재배포금지",
"video_id": "umbAs-igFhc"
} | 16 | [
"The current frame of the video shows a news anchor sitting at a desk in a news studio, delivering a weather report. Behind the anchor, there is a weather map showing a large, swirling storm system, indicating a typhoon. The text on the screen is in Korean and mentions \"Typhoon Nanmadol\" and its effects in Japan ... | umbAs-igFhc | The video depicts the aftermath of Typhoon Nanmadol, a powerful typhoon that hit Japan and its coastal regions, including Gyeongsang Province. The events and actions described in the frames show the response efforts of emergency responders, who are assessing damage, securing structures, and addressing the effects of the typhoon, including flooding, wind damage, and landslides. The video highlights the severity of the typhoon's impact, with high wind speeds, strong rainfall, and coastal flooding. The network's continuous coverage emphasizes the critical nature of the response efforts to mitigate the damage and protect the public. |
Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan. | [] | [] | [] | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%83%9C%ED%92%8D_%EB%82%9C%EB%A7%88%EB%8F%8C_(2022%EB%85%84)",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe query is focused on a specific storm event, Typhoon Nanmadol, which is a strong typhoon. Therefore, it is reasonable to extract events related to the storm, such as the storm's name and its impact. Additionally, the query's mention of the Philippines adds specificity to the storm's location, which should help to narrow down the search results to more relevant events.\n\nEVENTS:\n1. Typhoon Nanmadol\n2. Typhoon Josie\n3. Japan\n4. Philippines\n5. Typhoon",
"en_article_excerpt": "Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.",
"en_article_url": "https://en.wikipedia.org/wiki/Typhoon_Nanmadol_(2022)",
"event_category": "disasters",
"event_name": "typhoon_nanmadol",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nBased on my analysis, I consider the following events to be relevant to the given query: Typhoon Nanmadol, Super Typhoon Josie, and related typhoons, as these events are directly related to the query, especially considering its geographical location (Philippines). I also consider the fact that Typhoon Nanmadol was a significant storm, making it a suitable event to extract.\n\nEVENTS:\n1. Typhoon Nanmadol\n2. Typhoon Josie\n3. Super Typhoon Josie\n4. Typhoon Nanmadol affects Japan",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nGiven the query \"Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan\", the relevant events that could be extracted in a video are related to the impact and damage caused by the storm. Specifically, the events should focus on the aftermath and consequences of the storm, including destruction, flooding, landslides, and loss of life.\n\nEVENTS:\n1. Typhoon Nanmadol causing widespread destruction in Japan, with landslides and floods reported\n2. Rescue efforts underway to locate survivors\n3. Emergency declarations issued by the Japanese government to evacuate people from affected areas\n4. Floodwaters and landslides affecting the city of Fukuoka\n5. Rescue workers searching for survivors in the affected areas",
"video_URL": "http://youtube.com/watch?v=fxLzed6AI2Q",
"video_description": "제14호 태풍 난마돌이 일본 규슈 남부에 상륙했습니다. 일본 기상청은 어제 저녁 7시쯤 태풍이 규슈 남부 가고시마 시 부근에 상륙했다고 발표했습니다. 기상청은 이번 태풍이 규슈 서부 내륙을 따라 북상해 오늘 오전 규슈 북부에서 북동쪽으로 방향을 꺾어 이동할 것으로 예보했습니다. 태풍은 일본 열도를 관통해 오는 21일 오후 홋카이도 동쪽 해역으로 빠져나갈 전망입니다. 일본 기상청은 이번 태풍이 \"지금까지 경험한 적 없는 위험한 태풍\"으로 수십 년에 한 번 있을 만한 대규모 재해가 발생할 가능성이 있다며 각별한 대비를 당부했습니다. 오늘 낮까지 하루 동안 규슈 남부에 최고 500mm, 규슈 북부와 시코쿠에 최고 400mm의 많은 비가 예보돼 있습니다. 한편 가장 먼저 태풍 영향권에 들어선 일본 규슈 남부 지역을 중심으로 10여 명이 강풍에 넘어져 다치는 등 피해가 이어지고 있습니다. 규슈 지역 각 지자체는 주민들을 대상으로 안전한 곳으로 신속히 이동할 것을 권하는 '피난 지시'를 내리는 등 만일의 사태에 대비하고 있습니다. 일본 정부는 재해대책본부를 설치하고 태풍의 움직임과 주민 대피 상황, 그리고 피해 집계 등에 나섰습니다. 기시다 총리는 관계 각료회의를 소집해 국민에게 \"최신 기상 정보와 피난 정보에 주의해 신속히 생명을 지키는 행동을 해달라\"고 촉구했습니다. YTN 이경아 (kalee@ytn.co.kr) ※ '당신의 제보가 뉴스가 됩니다' [카카오톡] YTN 검색해 채널 추가 [전화] 02-398-8585 [메일] social@ytn.co.kr ▶ 기사 원문 : https://www.ytn.co.kr/_ln/0104_202209190251207843 ▶ 제보 하기 : https://mj.ytn.co.kr/mj/mj_write.php ▣ YTN 유튜브 채널 구독 : http://goo.gl/Ytb5SZ ⓒ YTN 무단 전재 및 재배포금지",
"video_id": "fxLzed6AI2Q"
} | 16 | [
"The image shows a news anchor presenting a weather report. The background features a stormy weather map with a focus on Japan. The text on the screen is in Korean and mentions a typhoon approaching Japan, with a category 5 storm affecting the region and warnings for heavy rain and strong winds.",
"The news ancho... | fxLzed6AI2Q | A typhoon named Nami-dol affected the region of Kyushu, Japan, causing severe damage to buildings, infrastructure, and the environment. The storm brought strong winds and heavy rain, leading to widespread destruction. The storm's impact was most significant in urban areas, with debris scattered and buildings damaged. The storm resulted in record-breaking winds and widespread damage to essential infrastructure. The extent of the damage was reinforced by the connection to the typhoon, identified as Nami-dol, and the impact on the area. |
Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan. | [] | [] | [] | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%83%9C%ED%92%8D_%EB%82%9C%EB%A7%88%EB%8F%8C_(2022%EB%85%84)",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe query is focused on a specific storm event, Typhoon Nanmadol, which is a strong typhoon. Therefore, it is reasonable to extract events related to the storm, such as the storm's name and its impact. Additionally, the query's mention of the Philippines adds specificity to the storm's location, which should help to narrow down the search results to more relevant events.\n\nEVENTS:\n1. Typhoon Nanmadol\n2. Typhoon Josie\n3. Japan\n4. Philippines\n5. Typhoon",
"en_article_excerpt": "Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.",
"en_article_url": "https://en.wikipedia.org/wiki/Typhoon_Nanmadol_(2022)",
"event_category": "disasters",
"event_name": "typhoon_nanmadol",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nBased on my analysis, I consider the following events to be relevant to the given query: Typhoon Nanmadol, Super Typhoon Josie, and related typhoons, as these events are directly related to the query, especially considering its geographical location (Philippines). I also consider the fact that Typhoon Nanmadol was a significant storm, making it a suitable event to extract.\n\nEVENTS:\n1. Typhoon Nanmadol\n2. Typhoon Josie\n3. Super Typhoon Josie\n4. Typhoon Nanmadol affects Japan",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nGiven the query \"Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan\", the relevant events that could be extracted in a video are related to the impact and damage caused by the storm. Specifically, the events should focus on the aftermath and consequences of the storm, including destruction, flooding, landslides, and loss of life.\n\nEVENTS:\n1. Typhoon Nanmadol causing widespread destruction in Japan, with landslides and floods reported\n2. Rescue efforts underway to locate survivors\n3. Emergency declarations issued by the Japanese government to evacuate people from affected areas\n4. Floodwaters and landslides affecting the city of Fukuoka\n5. Rescue workers searching for survivors in the affected areas",
"video_URL": "http://youtube.com/watch?v=tVUSMdbETuE",
"video_description": "오늘도 맑은 하늘 보긴 어렵겠습니다. 중국으로 향하는 태풍 '무이파'의 간접영향으로 오전까지 서쪽 지방에는 약한 비가 내리겠는데요, 비의 양이 적어서 낮 동안 활동하시는 데 큰 불편함은 없겠습니다. 때아닌 늦더위도 계속되겠습니다. 오늘 서울이 한낮에 28도, 광주가 31도까지 오르는 등 어제보다 높아 다소 덥겠습니다. 서울 등 수도권은 낮부터 비가 그치고 점차 갤 것으로 보입니다. 오늘 낮 기온은 예년보다 1∼4도가량 높겠고, 특히, 호남 지방을 중심으로 기온이 크게 오르겠습니다. 서울 28도, 전주 30도, 광주 31도가 예상됩니다. 다가오는 주말에는 서울 낮 기온이 30도까지 오르는 등 늦더위가 기승을 부리겠고요, 제주도와 동쪽 지방을 중심으로 비가 내릴 전망입니다. 한편, 오늘 오전 3시쯤, 일본 오키나와 동쪽 해상에서 제14호 태풍 '난마돌'이 발생했습니다. 이 태풍은 세력이 강해지면서 느리게 북상해 다음 주 월요일, 서귀포 남동쪽 약 280km 부근 해상까지 올라오겠습니다. 이후 진로는 유동적이지만, 일본 규슈 부근이나 대한해협으로 이동할 가능성이 있습니다. 앞으로 발표되는 태풍 정보를 잘 확인해주시기 바랍니다. 현재, 태풍 '무이파'의 간접 영향으로, 제주도와 남해안에는 '강풍주의보'가, 대부분 해상에는 '풍랑주의보'가 발효 중입니다. 내륙은 비가 대부분 그쳤지만 동해안과 제주도에는 비가 더 내릴 가능성 있는 만큼 시설물 관리와 해안가 안전사고에 각별히 유의하시기 바랍니다. 지금까지 YTN 원이다입니다. YTN 원이다 (wonleeda95@ytn.co.kr) ※ '당신의 제보가 뉴스가 됩니다' [카카오톡] YTN 검색해 채널 추가 [전화] 02-398-8585 [메일] social@ytn.co.kr ▶ 기사 원문 : https://www.ytn.co.kr/_ln/0108_202209141027402664 ▶ 제보 하기 : https://mj.ytn.co.kr/mj/mj_write.php ▣ YTN 유튜브 채널 구독 : http://goo.gl/Ytb5SZ ⓒ YTN 무단 전재 및 재배포금지",
"video_id": "tVUSMdbETuE"
} | 16 | [
"The image shows a female news reporter standing outdoors, holding a microphone with the \"YTN NEWS\" logo. She is reporting on the weather, indicating that there will be \"비 그치고 낮 동안 낮더위\" (after the rain, daytime heat) with an expected temperature of 31°C. The background features trees and a round structure, like... | tVUSMdbETuE | Here is a concise and accurate summary of the video:
The video depicts a series of frames showing a female news reporter reporting on the weather, specifically focusing on a typhoon (Typhoon 'Namdao') affecting the Korean Peninsula. The typhoon impacts include rain, late afternoon heat, and wind. The report is broadcast on a news channel and features a mix of shots of people in various activities, such as walking and riding bicycles, amidst the typhoon's impact on the urban and coastal areas. The video also includes weather forecasts and maps, highlighting the typhoon's path and expected impact until the afternoon. |
Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan. | [] | [] | [] | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%83%9C%ED%92%8D_%EB%82%9C%EB%A7%88%EB%8F%8C_(2022%EB%85%84)",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe query is focused on a specific storm event, Typhoon Nanmadol, which is a strong typhoon. Therefore, it is reasonable to extract events related to the storm, such as the storm's name and its impact. Additionally, the query's mention of the Philippines adds specificity to the storm's location, which should help to narrow down the search results to more relevant events.\n\nEVENTS:\n1. Typhoon Nanmadol\n2. Typhoon Josie\n3. Japan\n4. Philippines\n5. Typhoon",
"en_article_excerpt": "Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.",
"en_article_url": "https://en.wikipedia.org/wiki/Typhoon_Nanmadol_(2022)",
"event_category": "disasters",
"event_name": "typhoon_nanmadol",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nBased on my analysis, I consider the following events to be relevant to the given query: Typhoon Nanmadol, Super Typhoon Josie, and related typhoons, as these events are directly related to the query, especially considering its geographical location (Philippines). I also consider the fact that Typhoon Nanmadol was a significant storm, making it a suitable event to extract.\n\nEVENTS:\n1. Typhoon Nanmadol\n2. Typhoon Josie\n3. Super Typhoon Josie\n4. Typhoon Nanmadol affects Japan",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nGiven the query \"Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan\", the relevant events that could be extracted in a video are related to the impact and damage caused by the storm. Specifically, the events should focus on the aftermath and consequences of the storm, including destruction, flooding, landslides, and loss of life.\n\nEVENTS:\n1. Typhoon Nanmadol causing widespread destruction in Japan, with landslides and floods reported\n2. Rescue efforts underway to locate survivors\n3. Emergency declarations issued by the Japanese government to evacuate people from affected areas\n4. Floodwaters and landslides affecting the city of Fukuoka\n5. Rescue workers searching for survivors in the affected areas",
"video_URL": "http://youtube.com/watch?v=d6fZyZvF88k",
"video_description": "어젯밤 일본 규슈에 상륙한 태풍 난마돌의 영향으로 수십 명이 다치는 등 피해가 잇따르고 있습니다. NHK는 이번 태풍으로 현재까지 강풍에 넘어져 다치는 등 40여 명이 부상을 입은 것으로 집계됐다고 전했습니다. 이와 함께 태풍 영향권에 있는 지역 주민 약 960만 명에게는 신속한 안전 확보를 요청하는 최고 수준의 재해 경보와 피난 지시 등이 발령됐습니다. 순간최대풍속 50m/s에 이르는 강풍으로 공사용 대형 크레인이 꺾이고 지붕이 날아가는 등 물적 피해도 속출했습니다. 규슈 미야자키에는 지난 15일 이후 1,000mm에 이르는 비가 내리는 등 집중 호우로 인한 도로와 주택 침수도 잇따랐습니다. 많은 비로 하천 수위가 올라가 현재 30여 곳이 범람 위험 수위를 넘어선 것으로 나타났습니다. 일본 국토교통성은 댐 100여 곳의 수문을 미리 열어 범람을 막기 위해 방류를 실시했는데 이런 대규모 방류는 사상 처음입니다. 이와 함께 규슈부터 시코쿠 지역까지 약 31만 가구가 정전됐습니다. 항공과 선박, 철도 등 교통편도 태풍의 영향에서 벗어날 때까지 운행이 대부분 중단됩니다. 규슈 지역을 중심으로 큰 피해를 남긴 태풍 난마돌은 내일 일본 혼슈를 관통해 홋카이도 남쪽 바다로 빠져나갈 전망입니다. 일본 재해 당국은 많은 비가 내린 뒤 토사 붕괴 등 추가 피해가 발생할 가능성에 대비하며 태풍 움직임을 주시하고 있습니다. YTN 이경아 (kalee@ytn.co.kr) ※ '당신의 제보가 뉴스가 됩니다' [카카오톡] YTN 검색해 채널 추가 [전화] 02-398-8585 [메일] social@ytn.co.kr ▶ 기사 원문 : https://www.ytn.co.kr/_ln/0104_202209191047268893 ▶ 제보 하기 : https://mj.ytn.co.kr/mj/mj_write.php ▣ YTN 유튜브 채널 구독 : http://goo.gl/Ytb5SZ ⓒ YTN 무단 전재 및 재배포금지",
"video_id": "d6fZyZvF88k"
} | 16 | [
"The image depicts a scene of strong winds impacting palm trees, with the trees leaning dramatically in one direction. Text in the image indicates that this is due to Typhoon Nanmadol, which is causing significant impact, with over 900,000 households experiencing power outages. The weather conditions suggest heavy ... | d6fZyZvF88k | Typhoon Nanmadol (also known as Typhoon Lan) made landfall in Taiwan in 2018, bringing severe weather conditions including strong winds, heavy rain, and potential flooding, with over 900,000 households experiencing power outages and landslides. The typhoon's impact on the region was significant, causing damage to buildings, infrastructure, and daily life. |
Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan. | [] | [] | [] | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%83%9C%ED%92%8D_%EB%82%9C%EB%A7%88%EB%8F%8C_(2022%EB%85%84)",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe query is focused on a specific storm event, Typhoon Nanmadol, which is a strong typhoon. Therefore, it is reasonable to extract events related to the storm, such as the storm's name and its impact. Additionally, the query's mention of the Philippines adds specificity to the storm's location, which should help to narrow down the search results to more relevant events.\n\nEVENTS:\n1. Typhoon Nanmadol\n2. Typhoon Josie\n3. Japan\n4. Philippines\n5. Typhoon",
"en_article_excerpt": "Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.",
"en_article_url": "https://en.wikipedia.org/wiki/Typhoon_Nanmadol_(2022)",
"event_category": "disasters",
"event_name": "typhoon_nanmadol",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nBased on my analysis, I consider the following events to be relevant to the given query: Typhoon Nanmadol, Super Typhoon Josie, and related typhoons, as these events are directly related to the query, especially considering its geographical location (Philippines). I also consider the fact that Typhoon Nanmadol was a significant storm, making it a suitable event to extract.\n\nEVENTS:\n1. Typhoon Nanmadol\n2. Typhoon Josie\n3. Super Typhoon Josie\n4. Typhoon Nanmadol affects Japan",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nGiven the query \"Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan\", the relevant events that could be extracted in a video are related to the impact and damage caused by the storm. Specifically, the events should focus on the aftermath and consequences of the storm, including destruction, flooding, landslides, and loss of life.\n\nEVENTS:\n1. Typhoon Nanmadol causing widespread destruction in Japan, with landslides and floods reported\n2. Rescue efforts underway to locate survivors\n3. Emergency declarations issued by the Japanese government to evacuate people from affected areas\n4. Floodwaters and landslides affecting the city of Fukuoka\n5. Rescue workers searching for survivors in the affected areas",
"video_URL": "http://youtube.com/watch?v=ItDO3MSUQRQ",
"video_description": "힌남노의 상처가 채 가시기도 전에 이번에는 제 14호 태풍 '난마돌'이 북상하고 있습니다. 아직 경로는 유동적이지만 기상청은 다음주 월요일쯤 제주도를 향해 북진할 가능성이 있다고 보고 상황을 예의주시하고 있습니다. #가을태풍#태풍#태풍경로#힌남노#난마돌#태풍난마돌#제14호태풍#기상청#므르복#기상정보#이슈픽#자막뉴스 ☞MBN NEWS 구독하기 : https://www.youtube.com/channel/UCG9aFJTZ-lMCHAiO1KJsirg?sub_confirmation=1 ☞MPLAY 구독하기 : https://www.youtube.com/channel/UC84IlgY5HIzxKBZOt8XM5UA?sub_confirmation=1 ☞MBN Entertainment 구독하기 : https://www.youtube.com/channel/UCYyLIlOJyqkAFKlVjzX5img?sub_confirmation=1 ☞MBN 공식 홈페이지 : https://www.mbn.co.kr",
"video_id": "ItDO3MSUQRQ"
} | 16 | [
"The image shows a satellite view of weather patterns over East Asia, highlighting several large and intense cyclones. The text in the bottom left corner is in Korean, likely providing information about the weather system, including the presence of multiple typhoons or tropical storms. The area depicted includes pa... | ItDO3MSUQRQ | The video depicts a powerful typhoon series in East Asia, with intense cyclones, strong winds, and heavy rainfall affecting various locations. The typhoons are named 'Typhoon Merubok', 'Typhoon Namtheun', and 'Typhoon Namamudo', indicating a series of severe storms in the region.
The video focuses on the impact of these storms on various locations, including coastal areas, islands, and cities, emphasizing the need for preparedness and caution. The storms cause significant damage and disruption, with notable damage visible on the streets, including large rocks, debris, and empty roads.
The video also highlights the progression of the storms, with the storms moving northward off the western coast of Kyushu, Japan, and affecting coastal areas such as Jejudo Island. The storms cause hazardous conditions, with strong winds, heavy rainfall, and rough seas, leading to a focus on safety and caution.
The video notes that the storms are part of a larger weather forecast, with the Korea Meteorological Administration providing updates on the trajectory and potential impact of the storms. The overall context of the video emphasizes the importance of staying vigilant and taking precautions due to the severe weather conditions, with the storms potentially causing widespread damage and disruption. |
Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan. | [] | [] | [] | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%83%9C%ED%92%8D_%EB%82%9C%EB%A7%88%EB%8F%8C_(2022%EB%85%84)",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe query is focused on a specific storm event, Typhoon Nanmadol, which is a strong typhoon. Therefore, it is reasonable to extract events related to the storm, such as the storm's name and its impact. Additionally, the query's mention of the Philippines adds specificity to the storm's location, which should help to narrow down the search results to more relevant events.\n\nEVENTS:\n1. Typhoon Nanmadol\n2. Typhoon Josie\n3. Japan\n4. Philippines\n5. Typhoon",
"en_article_excerpt": "Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.",
"en_article_url": "https://en.wikipedia.org/wiki/Typhoon_Nanmadol_(2022)",
"event_category": "disasters",
"event_name": "typhoon_nanmadol",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nBased on my analysis, I consider the following events to be relevant to the given query: Typhoon Nanmadol, Super Typhoon Josie, and related typhoons, as these events are directly related to the query, especially considering its geographical location (Philippines). I also consider the fact that Typhoon Nanmadol was a significant storm, making it a suitable event to extract.\n\nEVENTS:\n1. Typhoon Nanmadol\n2. Typhoon Josie\n3. Super Typhoon Josie\n4. Typhoon Nanmadol affects Japan",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nGiven the query \"Typhoon Nanmadol, known in the Philippines as Super Typhoon Josie, was a powerful tropical cyclone that impacted Japan\", the relevant events that could be extracted in a video are related to the impact and damage caused by the storm. Specifically, the events should focus on the aftermath and consequences of the storm, including destruction, flooding, landslides, and loss of life.\n\nEVENTS:\n1. Typhoon Nanmadol causing widespread destruction in Japan, with landslides and floods reported\n2. Rescue efforts underway to locate survivors\n3. Emergency declarations issued by the Japanese government to evacuate people from affected areas\n4. Floodwaters and landslides affecting the city of Fukuoka\n5. Rescue workers searching for survivors in the affected areas",
"video_URL": "http://youtube.com/watch?v=ZuMemCoZIUY",
"video_description": "북상 중인 제14호 태풍 '난마돌'이 일본 규슈에 상륙할 것으로 보입니다. 기상청은 '난마돌'이 서귀포 남동쪽 먼바다까지 북상한 뒤 급격히 방향을 틀어 다음 주 월요일쯤 일본 규슈에 상륙할 것으로 내다봤습니다. 이후 일본 열도를 따라 계속 북동진할 것으로 보입니다. 따라서 우리나라에 큰 피해를 줄 가능성은 낮아졌지만, 제주도와 남해안, 영남 동해안은 태풍의 직간접적인 영향권에 들어 비바람이 강하게 불 것으로 보입니다. 기상청은 태풍의 진로가 아직은 유동적인 만큼 앞으로 발표되는 태풍 진로에 주의를 기울여 달라고 당부했습니다. YTN 김진두 (jdkim@ytn.co.kr) #태풍난마돌 #난마돌 #규슈 [YTN 사이언스 기사원문] https://science.ytn.co.kr/program/program_view.php?s_mcd=0082&s_hcd=&key=202209151243393568 [프로그램 제작 문의] legbiz@ytn.co.kr",
"video_id": "ZuMemCoZIUY"
} | 16 | [
"A news anchor is reporting on a segment titled \"Science Today.\" The screen behind her displays text in Korean, indicating a discussion about a large-scale seismic wave and its impact in Japan.",
"The news anchor continues to report on the \"Science Today\" segment, maintaining focus on the topic of a large-sca... | ZuMemCoZIUY | A news report discusses a significant wave event in Japan, specifically a 14-hour tsunami, with related coastal protection measures and aftermath in mind. The report is part of a segment titled "Science Today" on seismic waves, showing a coastal area with large waves crashing against rocky shores, followed by images of coastal structures and a stormy atmosphere. The report continues to emphasize the impact of the wave on palm trees and coastal defense systems. It appears to be focused on a specific event and its aftermath in Japan, with a tropical or coastal location, possibly including urban or military settings. |
On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas. | [] | [
"\"Large scale floods in [location] at unspecified time\"",
"\"South Korean capital flooded",
"\"Protest against flood\"",
"\"Protests or demonstrations in response to South Korea floods at [location]\"",
"\"Large-scale emergency responders or rescue efforts in [Location] during [time period].",
"\"South ... | [
"\"Seoul floods\"",
"Seoul during rainy season",
"\"Large scale floods in Gangnam\"",
"\"South Korean capital city\"",
"\"Flood in Seoul on a large scale at a specific time\"",
"Seoul flood",
"\"Seoul flood relief and damage assessment in response to recent severe rainstorms on [TIME] in Seoul, looking ... | {
"article_url": "https://ko.wikipedia.org/wiki/2022%EB%85%84_%ED%95%9C%EA%B5%AD_%EC%A4%91%EB%B6%80_%EC%A7%91%EC%A4%91%ED%98%B8%EC%9A%B0",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe query \"On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas\" suggests that we are dealing with an event that occurred in Seoul, South Korea, and has a clear geographical location. Given the specificity of the query and the mention of Seoul and Gangnam District, it makes sense to extract events that are closely related to these locations, likely indicating flooding. This suggests that we can extract events such as \"Rains in Seoul\", \"Flood in Gangnam\", \"Flood in Seoul\", \"Seoul residents affected by floods\", and \"Seoul authorities respond to floods\".\n\nEVENTS:\n1. Rains in Seoul\n2. Flood in Gangnam\n3. Flood in Seoul\n4. Seoul residents affected by floods\n5. Seoul authorities respond to floods",
"en_article_excerpt": "On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas.",
"en_article_url": "https://en.wikipedia.org/wiki/2022_South_Korean_floods",
"event_category": "disasters",
"event_name": "2022_floods",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION\nThe events that make sense to extract from the query are related to severe weather conditions and the capital city of South Korea. Given the mention of \"floods\" and \"capital of South Korea\", it seems logical to consider extreme weather events such as heavy rainfall or flash flooding, which could be visualized in a video.\n\nEVENTS\n\n1. Flash flooding in Seoul\n2. Heavy rainfall in Seoul\n3. Flood warning issued for Seoul\n4. Seoul experiencing severe weather\n5. Heavy rainfall causes widespread flooding",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nThe query \"large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas\" suggests that we need to find events that are visually depicted in a video of the area affected by the floods. Considering this context, we can expect to find images of flooded areas, damaged buildings, and possibly protesters or emergency responders responding to the disaster. Given the scale and location of the floods, we should expect to see visual evidence of extensive damage, destruction, and chaotic scenes, which are likely to be captured by the video.\n\nEVENTS:\n1. Flood damage to buildings\n2. Protests or demonstrations in response to the floods\n3. Emergency responders or rescue efforts\n4. Destruction of infrastructure (roads, bridges, etc.)\n5. Chaos or scenes of crisis and disorder in the streets",
"video_URL": "http://twitter.com/i/status/1556640821054787588",
"video_description": "I just found out about the subway floods in Korea, 모두 건강 유의하세요 ! 아무쪼록 인명피해는 없게 해주세요.",
"video_id": "1556640821054787588"
} | 16 | [
"Water is visibly dripping from the ceiling onto the floor in an indoor subway or train station. The station appears to be experiencing a leak, and there is a puddle forming on the ground. The area is marked with a blue and green sign, and there are informational posters and maps on the wall. The scene is likely pa... | 1556640821054787588 | A subway station is experiencing a severe water leakage, causing a significant disruption and hazard. The incident began with a dripping leak from the ceiling onto the floor in an indoor station, initially with a blue and green sign indicating a public transportation system. The leak worsened, forming a larger puddle and mist on the floor, posing a significant safety hazard. Over time, the intensity of the leak increased, with the water cascade and splash creating a substantial hazard. The situation remained unresolved with no visible efforts to address the leakage, indicating a critical and ongoing situation. |
On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas. | [] | [
"\"Large scale floods in [location] at unspecified time\"",
"\"South Korean capital flooded",
"\"Protest against flood\"",
"\"Protests or demonstrations in response to South Korea floods at [location]\"",
"\"Large-scale emergency responders or rescue efforts in [Location] during [time period].",
"\"South ... | [
"\"Seoul floods\"",
"Seoul during rainy season",
"\"Large scale floods in Gangnam\"",
"\"South Korean capital city\"",
"\"Flood in Seoul on a large scale at a specific time\"",
"Seoul flood",
"\"Seoul flood relief and damage assessment in response to recent severe rainstorms on [TIME] in Seoul, looking ... | {
"article_url": "https://ko.wikipedia.org/wiki/2022%EB%85%84_%ED%95%9C%EA%B5%AD_%EC%A4%91%EB%B6%80_%EC%A7%91%EC%A4%91%ED%98%B8%EC%9A%B0",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe query \"On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas\" suggests that we are dealing with an event that occurred in Seoul, South Korea, and has a clear geographical location. Given the specificity of the query and the mention of Seoul and Gangnam District, it makes sense to extract events that are closely related to these locations, likely indicating flooding. This suggests that we can extract events such as \"Rains in Seoul\", \"Flood in Gangnam\", \"Flood in Seoul\", \"Seoul residents affected by floods\", and \"Seoul authorities respond to floods\".\n\nEVENTS:\n1. Rains in Seoul\n2. Flood in Gangnam\n3. Flood in Seoul\n4. Seoul residents affected by floods\n5. Seoul authorities respond to floods",
"en_article_excerpt": "On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas.",
"en_article_url": "https://en.wikipedia.org/wiki/2022_South_Korean_floods",
"event_category": "disasters",
"event_name": "2022_floods",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION\nThe events that make sense to extract from the query are related to severe weather conditions and the capital city of South Korea. Given the mention of \"floods\" and \"capital of South Korea\", it seems logical to consider extreme weather events such as heavy rainfall or flash flooding, which could be visualized in a video.\n\nEVENTS\n\n1. Flash flooding in Seoul\n2. Heavy rainfall in Seoul\n3. Flood warning issued for Seoul\n4. Seoul experiencing severe weather\n5. Heavy rainfall causes widespread flooding",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nThe query \"large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas\" suggests that we need to find events that are visually depicted in a video of the area affected by the floods. Considering this context, we can expect to find images of flooded areas, damaged buildings, and possibly protesters or emergency responders responding to the disaster. Given the scale and location of the floods, we should expect to see visual evidence of extensive damage, destruction, and chaotic scenes, which are likely to be captured by the video.\n\nEVENTS:\n1. Flood damage to buildings\n2. Protests or demonstrations in response to the floods\n3. Emergency responders or rescue efforts\n4. Destruction of infrastructure (roads, bridges, etc.)\n5. Chaos or scenes of crisis and disorder in the streets",
"video_URL": "http://twitter.com/i/status/1556678909436932096",
"video_description": "This is absolutely insane video in South Korea. Extreme flooding that hasn’t been seen in 80 years or more. Very extreme flooding, causing massive problems in Seoul and the entirety of the country. #SouthKorea #flooding #weather",
"video_id": "1556678909436932096"
} | 16 | [
"The image shows a flooded street at night, with several buses partially submerged in water. The water level is high, reaching the lower parts of the buses. There are street signs and poles partially submerged, indicating the severity of the flooding. The text in Korean at the bottom of the image suggests that ther... | 1556678909436932096 | The video depicts a severe flood event in the city of Sadangyeok, where heavy rainfall causes widespread flooding, submerging buses and street furniture. The flood affects traffic, causing buses to be partially submerged, and disrupts daily activities. The floodwaters continue to rise, with the water level reaching high levels, causing inconvenience to residents and emergency responders. Despite efforts to mitigate the situation, the flood control and rescue efforts are insufficient, highlighting the severity of the issue and the need for urgent assistance. |
On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas. | [] | [
"\"Large scale floods in [location] at unspecified time\"",
"\"South Korean capital flooded",
"\"Protest against flood\"",
"\"Protests or demonstrations in response to South Korea floods at [location]\"",
"\"Large-scale emergency responders or rescue efforts in [Location] during [time period].",
"\"South ... | [
"\"Seoul floods\"",
"Seoul during rainy season",
"\"Large scale floods in Gangnam\"",
"\"South Korean capital city\"",
"\"Flood in Seoul on a large scale at a specific time\"",
"Seoul flood",
"\"Seoul flood relief and damage assessment in response to recent severe rainstorms on [TIME] in Seoul, looking ... | {
"article_url": "https://ko.wikipedia.org/wiki/2022%EB%85%84_%ED%95%9C%EA%B5%AD_%EC%A4%91%EB%B6%80_%EC%A7%91%EC%A4%91%ED%98%B8%EC%9A%B0",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe query \"On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas\" suggests that we are dealing with an event that occurred in Seoul, South Korea, and has a clear geographical location. Given the specificity of the query and the mention of Seoul and Gangnam District, it makes sense to extract events that are closely related to these locations, likely indicating flooding. This suggests that we can extract events such as \"Rains in Seoul\", \"Flood in Gangnam\", \"Flood in Seoul\", \"Seoul residents affected by floods\", and \"Seoul authorities respond to floods\".\n\nEVENTS:\n1. Rains in Seoul\n2. Flood in Gangnam\n3. Flood in Seoul\n4. Seoul residents affected by floods\n5. Seoul authorities respond to floods",
"en_article_excerpt": "On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas.",
"en_article_url": "https://en.wikipedia.org/wiki/2022_South_Korean_floods",
"event_category": "disasters",
"event_name": "2022_floods",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION\nThe events that make sense to extract from the query are related to severe weather conditions and the capital city of South Korea. Given the mention of \"floods\" and \"capital of South Korea\", it seems logical to consider extreme weather events such as heavy rainfall or flash flooding, which could be visualized in a video.\n\nEVENTS\n\n1. Flash flooding in Seoul\n2. Heavy rainfall in Seoul\n3. Flood warning issued for Seoul\n4. Seoul experiencing severe weather\n5. Heavy rainfall causes widespread flooding",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nThe query \"large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas\" suggests that we need to find events that are visually depicted in a video of the area affected by the floods. Considering this context, we can expect to find images of flooded areas, damaged buildings, and possibly protesters or emergency responders responding to the disaster. Given the scale and location of the floods, we should expect to see visual evidence of extensive damage, destruction, and chaotic scenes, which are likely to be captured by the video.\n\nEVENTS:\n1. Flood damage to buildings\n2. Protests or demonstrations in response to the floods\n3. Emergency responders or rescue efforts\n4. Destruction of infrastructure (roads, bridges, etc.)\n5. Chaos or scenes of crisis and disorder in the streets",
"video_URL": "http://twitter.com/i/status/1556973165556903939",
"video_description": "Visuals after yesterday's floods... Rains continue in S. Korea on Tuesday; \"level 3\" warning in operation to prevent damage. Be careful.. stay Safe.. #SouthKorea #flooding #flood #Seoul #climate #rain",
"video_id": "1556973165556903939"
} | 16 | [
"A Porsche car is driving through a flooded street during heavy rain, navigating around water puddles. A person wearing a raincoat stands near the car, possibly taking shelter or observing the situation. The street is lined with buildings, some with visible signs and shop fronts. The scene is illuminated by streetl... | 1556973165556903939 | The scene takes place in a busy urban environment, primarily focusing on a Korean-owned restaurant, "100년 영동 생굴백" (Hansin, Lifelong Living Home). The scene begins with a car driving through a flooded street during heavy rain, followed by a person skateboarding and then walking on a wet street, emphasizing the city's urban atmosphere. As the rain continues, the scene transitions to a bustling shopping area with neon signs and billboards illuminating the night. The atmosphere remains lively, with individuals navigating through the rain, including a woman walking with an umbrella, and people walking with umbrellas, despite the inclement weather. The scene gradually shifts to indoor locations, with a building entrance sign reading "100년 영동 생굴백" and a woman walking through a wet street, showcasing the restaurant's presence in the city. The rain continues to fall, reflecting off the wet pavement, and the scene becomes less crowded, with only a few individuals visible, emphasizing the city's busy and vibrant atmosphere. |
On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas. | [] | [
"\"Large scale floods in [location] at unspecified time\"",
"\"South Korean capital flooded",
"\"Protest against flood\"",
"\"Protests or demonstrations in response to South Korea floods at [location]\"",
"\"Large-scale emergency responders or rescue efforts in [Location] during [time period].",
"\"South ... | [
"\"Seoul floods\"",
"Seoul during rainy season",
"\"Large scale floods in Gangnam\"",
"\"South Korean capital city\"",
"\"Flood in Seoul on a large scale at a specific time\"",
"Seoul flood",
"\"Seoul flood relief and damage assessment in response to recent severe rainstorms on [TIME] in Seoul, looking ... | {
"article_url": "https://ko.wikipedia.org/wiki/2022%EB%85%84_%ED%95%9C%EA%B5%AD_%EC%A4%91%EB%B6%80_%EC%A7%91%EC%A4%91%ED%98%B8%EC%9A%B0",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe query \"On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas\" suggests that we are dealing with an event that occurred in Seoul, South Korea, and has a clear geographical location. Given the specificity of the query and the mention of Seoul and Gangnam District, it makes sense to extract events that are closely related to these locations, likely indicating flooding. This suggests that we can extract events such as \"Rains in Seoul\", \"Flood in Gangnam\", \"Flood in Seoul\", \"Seoul residents affected by floods\", and \"Seoul authorities respond to floods\".\n\nEVENTS:\n1. Rains in Seoul\n2. Flood in Gangnam\n3. Flood in Seoul\n4. Seoul residents affected by floods\n5. Seoul authorities respond to floods",
"en_article_excerpt": "On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas.",
"en_article_url": "https://en.wikipedia.org/wiki/2022_South_Korean_floods",
"event_category": "disasters",
"event_name": "2022_floods",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION\nThe events that make sense to extract from the query are related to severe weather conditions and the capital city of South Korea. Given the mention of \"floods\" and \"capital of South Korea\", it seems logical to consider extreme weather events such as heavy rainfall or flash flooding, which could be visualized in a video.\n\nEVENTS\n\n1. Flash flooding in Seoul\n2. Heavy rainfall in Seoul\n3. Flood warning issued for Seoul\n4. Seoul experiencing severe weather\n5. Heavy rainfall causes widespread flooding",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nThe query \"large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas\" suggests that we need to find events that are visually depicted in a video of the area affected by the floods. Considering this context, we can expect to find images of flooded areas, damaged buildings, and possibly protesters or emergency responders responding to the disaster. Given the scale and location of the floods, we should expect to see visual evidence of extensive damage, destruction, and chaotic scenes, which are likely to be captured by the video.\n\nEVENTS:\n1. Flood damage to buildings\n2. Protests or demonstrations in response to the floods\n3. Emergency responders or rescue efforts\n4. Destruction of infrastructure (roads, bridges, etc.)\n5. Chaos or scenes of crisis and disorder in the streets",
"video_URL": "http://twitter.com/i/status/1556659852054634496",
"video_description": "Dorimcheon stream in #Seoul's Gwanak District right now. People are now asking where President Yoon Suk-yeol is and what he's doing. Mayor Oh has just returned to office to deal with disaster, but it's not just Seoul. #SouthKorea #flooding #floodsituation",
"video_id": "1556659852054634496"
} | 16 | [
"The image depicts a flooded street at night, illuminated by streetlights. The water appears to be moving rapidly, suggesting strong currents. The scene is likely taken from a bridge or elevated position, as a railing is visible in the foreground. There are trees and possibly buildings in the background, partially ... | 1556659852054634496 | A severe flood causes widespread flooding in a urban area, characterized by chaotic and continuous heavy rain. The flood continues to affect the streets, with water levels rising and strong currents evident. The scene is framed from multiple angles, including elevated positions, which provide a clear view of the flooded areas and the movement of water. The events depicted show the community's efforts to navigate the flooded streets, including pedestrians trying to avoid flooded areas and bicycles parked on elevated platforms. The scene continues to show that the flood is persistent and has a significant impact on the environment, with trees and buildings partially submerged in the water. |
On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas. | [] | [
"\"Large scale floods in [location] at unspecified time\"",
"\"South Korean capital flooded",
"\"Protest against flood\"",
"\"Protests or demonstrations in response to South Korea floods at [location]\"",
"\"Large-scale emergency responders or rescue efforts in [Location] during [time period].",
"\"South ... | [
"\"Seoul floods\"",
"Seoul during rainy season",
"\"Large scale floods in Gangnam\"",
"\"South Korean capital city\"",
"\"Flood in Seoul on a large scale at a specific time\"",
"Seoul flood",
"\"Seoul flood relief and damage assessment in response to recent severe rainstorms on [TIME] in Seoul, looking ... | {
"article_url": "https://ko.wikipedia.org/wiki/2022%EB%85%84_%ED%95%9C%EA%B5%AD_%EC%A4%91%EB%B6%80_%EC%A7%91%EC%A4%91%ED%98%B8%EC%9A%B0",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe query \"On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas\" suggests that we are dealing with an event that occurred in Seoul, South Korea, and has a clear geographical location. Given the specificity of the query and the mention of Seoul and Gangnam District, it makes sense to extract events that are closely related to these locations, likely indicating flooding. This suggests that we can extract events such as \"Rains in Seoul\", \"Flood in Gangnam\", \"Flood in Seoul\", \"Seoul residents affected by floods\", and \"Seoul authorities respond to floods\".\n\nEVENTS:\n1. Rains in Seoul\n2. Flood in Gangnam\n3. Flood in Seoul\n4. Seoul residents affected by floods\n5. Seoul authorities respond to floods",
"en_article_excerpt": "On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas.",
"en_article_url": "https://en.wikipedia.org/wiki/2022_South_Korean_floods",
"event_category": "disasters",
"event_name": "2022_floods",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION\nThe events that make sense to extract from the query are related to severe weather conditions and the capital city of South Korea. Given the mention of \"floods\" and \"capital of South Korea\", it seems logical to consider extreme weather events such as heavy rainfall or flash flooding, which could be visualized in a video.\n\nEVENTS\n\n1. Flash flooding in Seoul\n2. Heavy rainfall in Seoul\n3. Flood warning issued for Seoul\n4. Seoul experiencing severe weather\n5. Heavy rainfall causes widespread flooding",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nThe query \"large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas\" suggests that we need to find events that are visually depicted in a video of the area affected by the floods. Considering this context, we can expect to find images of flooded areas, damaged buildings, and possibly protesters or emergency responders responding to the disaster. Given the scale and location of the floods, we should expect to see visual evidence of extensive damage, destruction, and chaotic scenes, which are likely to be captured by the video.\n\nEVENTS:\n1. Flood damage to buildings\n2. Protests or demonstrations in response to the floods\n3. Emergency responders or rescue efforts\n4. Destruction of infrastructure (roads, bridges, etc.)\n5. Chaos or scenes of crisis and disorder in the streets",
"video_URL": "http://twitter.com/i/status/1556995436052873218",
"video_description": "Heavy rain in Seoul floods train stations, submerges vehicles. #Seoul #flooding #floods #SouthKorea",
"video_id": "1556995436052873218"
} | 16 | [
"The image shows a flooded staircase with water rushing down the steps, creating a strong current. The stairs are enclosed by walls with handrails on both sides. The water appears muddy and is flowing vigorously, indicating a significant amount of water accumulation. The scene suggests a severe flooding situation, ... | 1556995436052873218 | A severe flood has occurred in an enclosed space, with water rushing down a flooded staircase in a powerful and relentless current. The scene is chaotic and intense, with the water cascading down the stairs in a forceful and turbulent manner. The flooding is severe and ongoing, with a significant amount of water accumulating in the space. The water's movement is forceful and overwhelming, creating a sense of urgency and danger. The scene suggests that the flooding is severe and persistent, indicating that the situation is critical and requires immediate attention. |
On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas. | [] | [
"\"Large scale floods in [location] at unspecified time\"",
"\"South Korean capital flooded",
"\"Protest against flood\"",
"\"Protests or demonstrations in response to South Korea floods at [location]\"",
"\"Large-scale emergency responders or rescue efforts in [Location] during [time period].",
"\"South ... | [
"\"Seoul floods\"",
"Seoul during rainy season",
"\"Large scale floods in Gangnam\"",
"\"South Korean capital city\"",
"\"Flood in Seoul on a large scale at a specific time\"",
"Seoul flood",
"\"Seoul flood relief and damage assessment in response to recent severe rainstorms on [TIME] in Seoul, looking ... | {
"article_url": "https://ko.wikipedia.org/wiki/2022%EB%85%84_%ED%95%9C%EA%B5%AD_%EC%A4%91%EB%B6%80_%EC%A7%91%EC%A4%91%ED%98%B8%EC%9A%B0",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe query \"On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas\" suggests that we are dealing with an event that occurred in Seoul, South Korea, and has a clear geographical location. Given the specificity of the query and the mention of Seoul and Gangnam District, it makes sense to extract events that are closely related to these locations, likely indicating flooding. This suggests that we can extract events such as \"Rains in Seoul\", \"Flood in Gangnam\", \"Flood in Seoul\", \"Seoul residents affected by floods\", and \"Seoul authorities respond to floods\".\n\nEVENTS:\n1. Rains in Seoul\n2. Flood in Gangnam\n3. Flood in Seoul\n4. Seoul residents affected by floods\n5. Seoul authorities respond to floods",
"en_article_excerpt": "On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas.",
"en_article_url": "https://en.wikipedia.org/wiki/2022_South_Korean_floods",
"event_category": "disasters",
"event_name": "2022_floods",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION\nThe events that make sense to extract from the query are related to severe weather conditions and the capital city of South Korea. Given the mention of \"floods\" and \"capital of South Korea\", it seems logical to consider extreme weather events such as heavy rainfall or flash flooding, which could be visualized in a video.\n\nEVENTS\n\n1. Flash flooding in Seoul\n2. Heavy rainfall in Seoul\n3. Flood warning issued for Seoul\n4. Seoul experiencing severe weather\n5. Heavy rainfall causes widespread flooding",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nThe query \"large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas\" suggests that we need to find events that are visually depicted in a video of the area affected by the floods. Considering this context, we can expect to find images of flooded areas, damaged buildings, and possibly protesters or emergency responders responding to the disaster. Given the scale and location of the floods, we should expect to see visual evidence of extensive damage, destruction, and chaotic scenes, which are likely to be captured by the video.\n\nEVENTS:\n1. Flood damage to buildings\n2. Protests or demonstrations in response to the floods\n3. Emergency responders or rescue efforts\n4. Destruction of infrastructure (roads, bridges, etc.)\n5. Chaos or scenes of crisis and disorder in the streets",
"video_URL": "http://twitter.com/i/status/1556833205671456768",
"video_description": "After heavy rain and #flashflood hit South Korea, water started coming inside the apartment as well. #Gangnam #Seoul #flooding #SouthKorea",
"video_id": "1556833205671456768"
} | 16 | [
"The frame shows an elevator lobby with a sign above the elevator doors displaying the number \"23.\" The elevator doors are wooden and have a brass frame. There is a marble wall on the right, and a small sign with Korean text at the bottom.",
"The elevator doors are closed, and the sign above still displays \"23... | 1556833205671456768 | The video depicts a scene of an elevator in a Korean language signage system, with the doors being closed at the B2 level, and the doors and signage indicating a specific location: "서초아크로비스타". This suggests the elevator is likely taking passengers to the B2 floor of a building in Seoul, South Korea. |
On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas. | [] | [
"\"Large scale floods in [location] at unspecified time\"",
"\"South Korean capital flooded",
"\"Protest against flood\"",
"\"Protests or demonstrations in response to South Korea floods at [location]\"",
"\"Large-scale emergency responders or rescue efforts in [Location] during [time period].",
"\"South ... | [
"\"Seoul floods\"",
"Seoul during rainy season",
"\"Large scale floods in Gangnam\"",
"\"South Korean capital city\"",
"\"Flood in Seoul on a large scale at a specific time\"",
"Seoul flood",
"\"Seoul flood relief and damage assessment in response to recent severe rainstorms on [TIME] in Seoul, looking ... | {
"article_url": "https://ko.wikipedia.org/wiki/2022%EB%85%84_%ED%95%9C%EA%B5%AD_%EC%A4%91%EB%B6%80_%EC%A7%91%EC%A4%91%ED%98%B8%EC%9A%B0",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nThe query \"On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas\" suggests that we are dealing with an event that occurred in Seoul, South Korea, and has a clear geographical location. Given the specificity of the query and the mention of Seoul and Gangnam District, it makes sense to extract events that are closely related to these locations, likely indicating flooding. This suggests that we can extract events such as \"Rains in Seoul\", \"Flood in Gangnam\", \"Flood in Seoul\", \"Seoul residents affected by floods\", and \"Seoul authorities respond to floods\".\n\nEVENTS:\n1. Rains in Seoul\n2. Flood in Gangnam\n3. Flood in Seoul\n4. Seoul residents affected by floods\n5. Seoul authorities respond to floods",
"en_article_excerpt": "On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas.",
"en_article_url": "https://en.wikipedia.org/wiki/2022_South_Korean_floods",
"event_category": "disasters",
"event_name": "2022_floods",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION\nThe events that make sense to extract from the query are related to severe weather conditions and the capital city of South Korea. Given the mention of \"floods\" and \"capital of South Korea\", it seems logical to consider extreme weather events such as heavy rainfall or flash flooding, which could be visualized in a video.\n\nEVENTS\n\n1. Flash flooding in Seoul\n2. Heavy rainfall in Seoul\n3. Flood warning issued for Seoul\n4. Seoul experiencing severe weather\n5. Heavy rainfall causes widespread flooding",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 8 August 2022, large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nThe query \"large scale floods hit the capital of South Korea, Seoul, especially Gangnam District and surrounding areas\" suggests that we need to find events that are visually depicted in a video of the area affected by the floods. Considering this context, we can expect to find images of flooded areas, damaged buildings, and possibly protesters or emergency responders responding to the disaster. Given the scale and location of the floods, we should expect to see visual evidence of extensive damage, destruction, and chaotic scenes, which are likely to be captured by the video.\n\nEVENTS:\n1. Flood damage to buildings\n2. Protests or demonstrations in response to the floods\n3. Emergency responders or rescue efforts\n4. Destruction of infrastructure (roads, bridges, etc.)\n5. Chaos or scenes of crisis and disorder in the streets",
"video_URL": "http://twitter.com/i/status/1557013913216352258",
"video_description": "Summers in South Korea be like... #southkorea #Monsoon2022 #monsoonseason #summer2022 #flooding #summersouthkorea #여름 #침수 #summertime",
"video_id": "1557013913216352258"
} | 16 | [
"A person in a white shirt is holding an umbrella and appears to be shielding themselves from rain. Another person in a red shirt is bending over on the street, possibly picking something up. A bus is visible in the background with text in Korean, and an orange traffic cone is also present.",
"In the current fram... | 1557013913216352258 | A heavy rainstorm had a significant impact on a busy city street, with numerous vehicles and pedestrians navigating through flooded sections. The storm caused the water level to rise, reaching mid-calf height in some areas, and traffic management measures were put in place to facilitate safe passage. The scene followed a person navigating through flooded streets, with vehicles and pedestrians, as well as buildings with neon signs, indicating commercial activity. |
A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10. | [
"construction site major Korean cities",
"\"Korean construction site fires\"",
"Construction site fire at a major Korean building",
"\"Impact of fires on South Korea's infrastructure and public safety\"",
"\"South Korean authorities respond to injury in South Korea, at a time that is not specified.\"",
"\... | [
"fire construction site",
"\"construction site near me\"",
"construction site fire",
"Major fire on a specific location at a specific time",
"\"people injured\"",
"\"49 dead, 17 injured in 2023 mass shooting in [location]",
"\"Fire at [location] on [time]\"",
"\"injured person in [location] at unspeci... | [
"Construction site fires - major",
"\"construction site fire for injured in [location]\"",
"Construction site fire near [location] on [time]",
"\"major fire in any location\"",
"I couldn't find any available information that combines 'people killed' with the provided details.",
"people killed",
"\"major... | {
"article_url": "https://ko.wikipedia.org/wiki/%EC%9D%B4%EC%B2%9C_%EB%AC%BC%EB%A5%98%EC%84%BC%ED%84%B0_%EA%B3%B5%EC%82%AC%EC%9E%A5_%ED%99%94%EC%9E%AC",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION: \nExtracting events from the query that could be visually represented in a video is not possible because the query contains specific dates and locations that are difficult to visualize. This query seems to be a news event rather than a video, and it requires a specific date and location.\n\nHowever, it seems logical to extract events related to the query, as they are likely to be of interest to a video viewer.\n\nEVENTS:\n1. Construction site fire\n2. People killed\n3. Injuries\n4. South Korea",
"en_article_excerpt": "A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.",
"en_article_url": "https://en.wikipedia.org/wiki/2020_Icheon_fire",
"event_category": "disasters",
"event_name": "icheon_fire",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThe given query contains vague language about a significant event, \"a major fire,\" which could be interpreted in different ways. To extract concrete events related to the query, we need to narrow down the context. Since the event occurred in Icheon, South Korea, we will only consider events that are directly related to that location. Furthermore, given the date \"29 April 2020,\" we need to find events that are relevant to a major fire occurring in that time frame.\n\nEVENTS:\n1. Construction site fire at a major Korean building \n2. South Korean authorities respond to the fire \n3. Firefighters attempt to extinguish the blaze \n4. Rescue efforts underway for injured workers \n5. Emergency services declare the scene",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\n\nThe given query describes a tragic event involving a fire that occurred at a construction site in Icheon, South Korea. This event is highly relevant for a video search query as it involves a specific location and a sense of tragedy and loss.\n\nEVENTS:\n\n1. Construction site fire\n2. Human casualties (39 people killed)\n3. Injury to another person",
"video_URL": "http://youtube.com/watch?v=SCkWqQ49yUQ",
"video_description": "이천 물류센터 화재와 관련해 관할 지자체가 재난안전대책본부를 구성했습니다. 경기 이천시는 엄태준 시장을 본부장으로 30명 규모의 재난안전대책본부를 편성했다고 밝혔습니다. 의료구호반과 장비지원반 등 11개 반으로 꾸려진 대책본부는 환자 이송과 현장 수색 작업 등에 참여합니다. 신준명[shinjm7529@ytn.co.kr] ※ '당신의 제보가 뉴스가 됩니다' YTN은 여러분의 소중한 제보를 기다립니다. [카카오톡] YTN을 검색해 채널 추가 [전화] 02-398-8585 [메일] social@ytn.co.kr [온라인 제보] www.ytn.co.kr ▶ 기사 원문 : https://www.ytn.co.kr/_ln/0103_202004292044009865 ▶ 제보 안내 : http://goo.gl/gEvsAL, 모바일앱, social@ytn.co.kr, #2424 ▣ YTN 유튜브 채널 구독 : http://goo.gl/Ytb5SZ ⓒ YTN & YTN PLUS 무단 전재 및 재배포금지",
"video_id": "SCkWqQ49yUQ"
} | 16 | [
"The image depicts a large building that has been severely damaged by fire, with visible smoke and burn marks on its exterior. Emergency personnel, including firefighters and medical staff, are present on the scene, indicating an ongoing response to the incident. Several emergency vehicles, such as fire trucks and ... | SCkWqQ49yUQ | A large building with severe fire damage has been reported in a major city, resulting in extensive fire damage to its facade and interior. Emergency personnel, including firefighters and medical responders, are present on the scene, coordinating their efforts to manage the situation and ensure the safety of the public. An ongoing logistics center has also been damaged, with melted and charred surfaces, and emergency services are working to assess the damage and initiate reconstruction efforts.
Multiple emergency vehicles, including fire trucks and ambulances, are parked in front of the building, indicating a coordinated response to the incident. The city has established an emergency response team to manage the situation and oversee reconstruction efforts. The city government has taken steps to address the disaster, including the establishment of the emergency response team and the restoration of essential services.
A logistics center, which is a critical infrastructure serving as a hub for the city's logistics operations, has been severely damaged in the fire. The destruction of the logistics center highlights the severity of the fire and the need for immediate attention to rebuild and restore essential services. The ongoing efforts of emergency responders, government officials, and construction teams aim to mitigate the impact of the disaster and promote recovery and reconstruction in the affected area. |
A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10. | [
"construction site major Korean cities",
"\"Korean construction site fires\"",
"Construction site fire at a major Korean building",
"\"Impact of fires on South Korea's infrastructure and public safety\"",
"\"South Korean authorities respond to injury in South Korea, at a time that is not specified.\"",
"\... | [
"fire construction site",
"\"construction site near me\"",
"construction site fire",
"Major fire on a specific location at a specific time",
"\"people injured\"",
"\"49 dead, 17 injured in 2023 mass shooting in [location]",
"\"Fire at [location] on [time]\"",
"\"injured person in [location] at unspeci... | [
"Construction site fires - major",
"\"construction site fire for injured in [location]\"",
"Construction site fire near [location] on [time]",
"\"major fire in any location\"",
"I couldn't find any available information that combines 'people killed' with the provided details.",
"people killed",
"\"major... | {
"article_url": "https://ko.wikipedia.org/wiki/%EC%9D%B4%EC%B2%9C_%EB%AC%BC%EB%A5%98%EC%84%BC%ED%84%B0_%EA%B3%B5%EC%82%AC%EC%9E%A5_%ED%99%94%EC%9E%AC",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION: \nExtracting events from the query that could be visually represented in a video is not possible because the query contains specific dates and locations that are difficult to visualize. This query seems to be a news event rather than a video, and it requires a specific date and location.\n\nHowever, it seems logical to extract events related to the query, as they are likely to be of interest to a video viewer.\n\nEVENTS:\n1. Construction site fire\n2. People killed\n3. Injuries\n4. South Korea",
"en_article_excerpt": "A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.",
"en_article_url": "https://en.wikipedia.org/wiki/2020_Icheon_fire",
"event_category": "disasters",
"event_name": "icheon_fire",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThe given query contains vague language about a significant event, \"a major fire,\" which could be interpreted in different ways. To extract concrete events related to the query, we need to narrow down the context. Since the event occurred in Icheon, South Korea, we will only consider events that are directly related to that location. Furthermore, given the date \"29 April 2020,\" we need to find events that are relevant to a major fire occurring in that time frame.\n\nEVENTS:\n1. Construction site fire at a major Korean building \n2. South Korean authorities respond to the fire \n3. Firefighters attempt to extinguish the blaze \n4. Rescue efforts underway for injured workers \n5. Emergency services declare the scene",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\n\nThe given query describes a tragic event involving a fire that occurred at a construction site in Icheon, South Korea. This event is highly relevant for a video search query as it involves a specific location and a sense of tragedy and loss.\n\nEVENTS:\n\n1. Construction site fire\n2. Human casualties (39 people killed)\n3. Injury to another person",
"video_URL": "http://youtube.com/watch?v=oRy1PP3KzDs",
"video_description": "29일 오후 경기도 이천 물류창고 공사현장에서 화재가 발생해 큰 인명피해를 냈습니다. 오후 7시30분 현재 확인된 사망자만 36명입니다.",
"video_id": "oRy1PP3KzDs"
} | 16 | [
"A building is on fire, with thick black smoke billowing out from multiple windows. Firefighters are spraying water on the structure to try to extinguish the flames. The text at the bottom of the image indicates that the fire occurred on the 29th at a public sales center.",
"A building is on fire, with dense blac... | oRy1PP3KzDs | A building caught fire at a public sales center on the 29th, with firefighters spraying water to combat the blaze. The fire started due to a high-temperature resistance test on the 4th floor and resulted in an estimated 78 fatalities. Firefighters continue to work to control the fire and locate missing individuals, with a total of 25 fatalities being reported. |
A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10. | [
"construction site major Korean cities",
"\"Korean construction site fires\"",
"Construction site fire at a major Korean building",
"\"Impact of fires on South Korea's infrastructure and public safety\"",
"\"South Korean authorities respond to injury in South Korea, at a time that is not specified.\"",
"\... | [
"fire construction site",
"\"construction site near me\"",
"construction site fire",
"Major fire on a specific location at a specific time",
"\"people injured\"",
"\"49 dead, 17 injured in 2023 mass shooting in [location]",
"\"Fire at [location] on [time]\"",
"\"injured person in [location] at unspeci... | [
"Construction site fires - major",
"\"construction site fire for injured in [location]\"",
"Construction site fire near [location] on [time]",
"\"major fire in any location\"",
"I couldn't find any available information that combines 'people killed' with the provided details.",
"people killed",
"\"major... | {
"article_url": "https://ko.wikipedia.org/wiki/%EC%9D%B4%EC%B2%9C_%EB%AC%BC%EB%A5%98%EC%84%BC%ED%84%B0_%EA%B3%B5%EC%82%AC%EC%9E%A5_%ED%99%94%EC%9E%AC",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION: \nExtracting events from the query that could be visually represented in a video is not possible because the query contains specific dates and locations that are difficult to visualize. This query seems to be a news event rather than a video, and it requires a specific date and location.\n\nHowever, it seems logical to extract events related to the query, as they are likely to be of interest to a video viewer.\n\nEVENTS:\n1. Construction site fire\n2. People killed\n3. Injuries\n4. South Korea",
"en_article_excerpt": "A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.",
"en_article_url": "https://en.wikipedia.org/wiki/2020_Icheon_fire",
"event_category": "disasters",
"event_name": "icheon_fire",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThe given query contains vague language about a significant event, \"a major fire,\" which could be interpreted in different ways. To extract concrete events related to the query, we need to narrow down the context. Since the event occurred in Icheon, South Korea, we will only consider events that are directly related to that location. Furthermore, given the date \"29 April 2020,\" we need to find events that are relevant to a major fire occurring in that time frame.\n\nEVENTS:\n1. Construction site fire at a major Korean building \n2. South Korean authorities respond to the fire \n3. Firefighters attempt to extinguish the blaze \n4. Rescue efforts underway for injured workers \n5. Emergency services declare the scene",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\n\nThe given query describes a tragic event involving a fire that occurred at a construction site in Icheon, South Korea. This event is highly relevant for a video search query as it involves a specific location and a sense of tragedy and loss.\n\nEVENTS:\n\n1. Construction site fire\n2. Human casualties (39 people killed)\n3. Injury to another person",
"video_URL": "http://youtube.com/watch?v=rqRqdg-oxkU",
"video_description": "이번에도 샌드위치 패널... 처참한 이천 물류센터 화재현장",
"video_id": "rqRqdg-oxkU"
} | 16 | [
"The image depicts a large industrial building engulfed in smoke, with visible flames and thick black smoke billowing from its upper levels. Firefighters are actively working to control the blaze, with water streams directed towards the building from fire trucks positioned below. The surrounding area shows a few in... | rqRqdg-oxkU | A large fire engulfed a large industrial building, with firefighters working to control the blaze using water from fire trucks below. The fire was still active, with thick smoke billowing from the building, and firefighters directed water streams towards the flames from fire trucks positioned below. The scene depicted a factory or warehouse setting, with industrial containers and a blue tent visible in the vicinity. |
A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10. | [
"construction site major Korean cities",
"\"Korean construction site fires\"",
"Construction site fire at a major Korean building",
"\"Impact of fires on South Korea's infrastructure and public safety\"",
"\"South Korean authorities respond to injury in South Korea, at a time that is not specified.\"",
"\... | [
"fire construction site",
"\"construction site near me\"",
"construction site fire",
"Major fire on a specific location at a specific time",
"\"people injured\"",
"\"49 dead, 17 injured in 2023 mass shooting in [location]",
"\"Fire at [location] on [time]\"",
"\"injured person in [location] at unspeci... | [
"Construction site fires - major",
"\"construction site fire for injured in [location]\"",
"Construction site fire near [location] on [time]",
"\"major fire in any location\"",
"I couldn't find any available information that combines 'people killed' with the provided details.",
"people killed",
"\"major... | {
"article_url": "https://ko.wikipedia.org/wiki/%EC%9D%B4%EC%B2%9C_%EB%AC%BC%EB%A5%98%EC%84%BC%ED%84%B0_%EA%B3%B5%EC%82%AC%EC%9E%A5_%ED%99%94%EC%9E%AC",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION: \nExtracting events from the query that could be visually represented in a video is not possible because the query contains specific dates and locations that are difficult to visualize. This query seems to be a news event rather than a video, and it requires a specific date and location.\n\nHowever, it seems logical to extract events related to the query, as they are likely to be of interest to a video viewer.\n\nEVENTS:\n1. Construction site fire\n2. People killed\n3. Injuries\n4. South Korea",
"en_article_excerpt": "A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.",
"en_article_url": "https://en.wikipedia.org/wiki/2020_Icheon_fire",
"event_category": "disasters",
"event_name": "icheon_fire",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThe given query contains vague language about a significant event, \"a major fire,\" which could be interpreted in different ways. To extract concrete events related to the query, we need to narrow down the context. Since the event occurred in Icheon, South Korea, we will only consider events that are directly related to that location. Furthermore, given the date \"29 April 2020,\" we need to find events that are relevant to a major fire occurring in that time frame.\n\nEVENTS:\n1. Construction site fire at a major Korean building \n2. South Korean authorities respond to the fire \n3. Firefighters attempt to extinguish the blaze \n4. Rescue efforts underway for injured workers \n5. Emergency services declare the scene",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\n\nThe given query describes a tragic event involving a fire that occurred at a construction site in Icheon, South Korea. This event is highly relevant for a video search query as it involves a specific location and a sense of tragedy and loss.\n\nEVENTS:\n\n1. Construction site fire\n2. Human casualties (39 people killed)\n3. Injury to another person",
"video_URL": "http://youtube.com/watch?v=6Z2tI74NGGQ",
"video_description": "[당시 현장 목격자 : 고개 돌리는 순간 연기가 엄청 심하게 나오더라고요. 설마 저게 불일까 그랬는데, 폭발음과 함께 불꽃이 일기 시작한 거죠.] 삽시간에 불길 번져 인명 피해 잇따라 \"우레탄 작업 중 폭발\" 추정 [서승현 / 경기 이천소방서장 : 화상 당하신 분들이 옷이 다 탄 걸로 봐서는 유증기가 폭발하면서 일어난 것으로 추정….] 완공 2개월 앞두고 참사 또 발생한 대형 화재 사상자 속출 ▶ 기사 원문 : https://www.ytn.co.kr/_ln/0103_202004292050572443 ▶ 제보 안내 : http://goo.gl/gEvsAL, 모바일앱, social@ytn.co.kr, #2424 ▣ YTN 유튜브 채널 구독 : http://goo.gl/Ytb5SZ ⓒ YTN & YTN PLUS 무단 전재 및 재배포금지",
"video_id": "6Z2tI74NGGQ"
} | 16 | [
"The image shows a large building engulfed in thick black smoke, indicating a significant fire. The building is surrounded by roads and some greenery. Text in the image reads \"이천 물류창고에서 화재\" (Translation: Fire at Icheon logistics warehouse), suggesting that the event is a warehouse fire in Icheon.",
"The image c... | 6Z2tI74NGGQ | A warehouse fire in Icheon, South Korea has resulted in significant damage to a large building and potential casualties. The fire started in the logistics warehouse and has been burning for several days. Firefighters and emergency responders, including a red fire engine, are actively working to contain and extinguish the blaze.
The fire has caused extensive damage to the building, with thick black smoke and flames visible from a distance. A fire truck and other emergency responders are present, and a person in yellow is seen wearing a helmet, likely a firefighter. The incident has resulted in concerns about potential casualties and the firefighters' efforts to manage the situation. |
A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10. | [
"construction site major Korean cities",
"\"Korean construction site fires\"",
"Construction site fire at a major Korean building",
"\"Impact of fires on South Korea's infrastructure and public safety\"",
"\"South Korean authorities respond to injury in South Korea, at a time that is not specified.\"",
"\... | [
"fire construction site",
"\"construction site near me\"",
"construction site fire",
"Major fire on a specific location at a specific time",
"\"people injured\"",
"\"49 dead, 17 injured in 2023 mass shooting in [location]",
"\"Fire at [location] on [time]\"",
"\"injured person in [location] at unspeci... | [
"Construction site fires - major",
"\"construction site fire for injured in [location]\"",
"Construction site fire near [location] on [time]",
"\"major fire in any location\"",
"I couldn't find any available information that combines 'people killed' with the provided details.",
"people killed",
"\"major... | {
"article_url": "https://ko.wikipedia.org/wiki/%EC%9D%B4%EC%B2%9C_%EB%AC%BC%EB%A5%98%EC%84%BC%ED%84%B0_%EA%B3%B5%EC%82%AC%EC%9E%A5_%ED%99%94%EC%9E%AC",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION: \nExtracting events from the query that could be visually represented in a video is not possible because the query contains specific dates and locations that are difficult to visualize. This query seems to be a news event rather than a video, and it requires a specific date and location.\n\nHowever, it seems logical to extract events related to the query, as they are likely to be of interest to a video viewer.\n\nEVENTS:\n1. Construction site fire\n2. People killed\n3. Injuries\n4. South Korea",
"en_article_excerpt": "A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.",
"en_article_url": "https://en.wikipedia.org/wiki/2020_Icheon_fire",
"event_category": "disasters",
"event_name": "icheon_fire",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThe given query contains vague language about a significant event, \"a major fire,\" which could be interpreted in different ways. To extract concrete events related to the query, we need to narrow down the context. Since the event occurred in Icheon, South Korea, we will only consider events that are directly related to that location. Furthermore, given the date \"29 April 2020,\" we need to find events that are relevant to a major fire occurring in that time frame.\n\nEVENTS:\n1. Construction site fire at a major Korean building \n2. South Korean authorities respond to the fire \n3. Firefighters attempt to extinguish the blaze \n4. Rescue efforts underway for injured workers \n5. Emergency services declare the scene",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\n\nThe given query describes a tragic event involving a fire that occurred at a construction site in Icheon, South Korea. This event is highly relevant for a video search query as it involves a specific location and a sense of tragedy and loss.\n\nEVENTS:\n\n1. Construction site fire\n2. Human casualties (39 people killed)\n3. Injury to another person",
"video_URL": "http://youtube.com/watch?v=JCKK0gu12zI",
"video_description": "경기 이천시 물류센터 화재가 대응 2단계로 격상한 가운데 인명피해가 속출하고 있다. 29일 경기도소방재난본부에 따르면 이날 오후 1시 32분 이천시 모가면 소고리 연면적 약 1만1043㎡인 물류센터 신축 공사현장에서 화재가 발생해 11명이 인명피해를 입었다.",
"video_id": "JCKK0gu12zI"
} | 16 | [
"The image shows an aerial view of a building surrounded by smoke. The text indicates that the building is a \"fuel cell factory\" experiencing an explosion or fire.",
"The image shows an aerial view of a factory area with thick black smoke billowing from a building. The text indicates that this is a \"fuel cell ... | JCKK0gu12zI | A fuel cell factory in South Korea experiences a catastrophic explosion, resulting in an ongoing emergency situation. A multi-branch fire is raging with thick smoke billowing from the building, supported by active firefighting efforts. Emergency services are engaging in their efforts to control and extinguish the blaze, with firefighters directing water from hoses towards the fire. |
A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10. | [
"construction site major Korean cities",
"\"Korean construction site fires\"",
"Construction site fire at a major Korean building",
"\"Impact of fires on South Korea's infrastructure and public safety\"",
"\"South Korean authorities respond to injury in South Korea, at a time that is not specified.\"",
"\... | [
"fire construction site",
"\"construction site near me\"",
"construction site fire",
"Major fire on a specific location at a specific time",
"\"people injured\"",
"\"49 dead, 17 injured in 2023 mass shooting in [location]",
"\"Fire at [location] on [time]\"",
"\"injured person in [location] at unspeci... | [
"Construction site fires - major",
"\"construction site fire for injured in [location]\"",
"Construction site fire near [location] on [time]",
"\"major fire in any location\"",
"I couldn't find any available information that combines 'people killed' with the provided details.",
"people killed",
"\"major... | {
"article_url": "https://ko.wikipedia.org/wiki/%EC%9D%B4%EC%B2%9C_%EB%AC%BC%EB%A5%98%EC%84%BC%ED%84%B0_%EA%B3%B5%EC%82%AC%EC%9E%A5_%ED%99%94%EC%9E%AC",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION: \nExtracting events from the query that could be visually represented in a video is not possible because the query contains specific dates and locations that are difficult to visualize. This query seems to be a news event rather than a video, and it requires a specific date and location.\n\nHowever, it seems logical to extract events related to the query, as they are likely to be of interest to a video viewer.\n\nEVENTS:\n1. Construction site fire\n2. People killed\n3. Injuries\n4. South Korea",
"en_article_excerpt": "A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.",
"en_article_url": "https://en.wikipedia.org/wiki/2020_Icheon_fire",
"event_category": "disasters",
"event_name": "icheon_fire",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThe given query contains vague language about a significant event, \"a major fire,\" which could be interpreted in different ways. To extract concrete events related to the query, we need to narrow down the context. Since the event occurred in Icheon, South Korea, we will only consider events that are directly related to that location. Furthermore, given the date \"29 April 2020,\" we need to find events that are relevant to a major fire occurring in that time frame.\n\nEVENTS:\n1. Construction site fire at a major Korean building \n2. South Korean authorities respond to the fire \n3. Firefighters attempt to extinguish the blaze \n4. Rescue efforts underway for injured workers \n5. Emergency services declare the scene",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\n\nThe given query describes a tragic event involving a fire that occurred at a construction site in Icheon, South Korea. This event is highly relevant for a video search query as it involves a specific location and a sense of tragedy and loss.\n\nEVENTS:\n\n1. Construction site fire\n2. Human casualties (39 people killed)\n3. Injury to another person",
"video_URL": "http://youtube.com/watch?v=ZFWsFwpfGlE",
"video_description": "경찰 오늘 이천 물류창고 화재 중간 수사결과 발표 [전국네트워크] 경기남부지방경찰청이 오늘 오전 38명의 목숨을 앗아간 '경기도 이천 물류창고 화재' 사고에 대한 중간 수사 결과를 발표합니다. 지난 4월 29일 경기 이천시의 한 물류창고 공사현장에서 불이 나 38명이 숨지고 10명이 다친 바 있는데, 화재 원인은 지하 2층 우레탄 작업 중 발생한 미상의 화원에 의한 폭발로 추정되고 있습니다. 경찰은 4차례의 대규모 현장 감식과 함께 화재와 관련된 80여 명을 조사해 이 가운데 17명에 대해 업무상 과실치사상 혐의 등을 적용해 입건했습니다. ☞ MBN 유튜브 구독하기 ☞ https://goo.gl/6ZsJGT 📢 MBN 유튜브 커뮤니티https://www.youtube.com/user/mbn/community?disable_polymer=1 MBN 페이스북 http://www.facebook.com/mbntv MBN 인스타그램 https://www.instagram.com/mbn_news/ #MBN뉴스#전국네트워크#이천물류창고",
"video_id": "ZFWsFwpfGlE"
} | 16 | [
"A news anchor is standing in front of a backdrop displaying a map with Korean text. The text indicates the segment is about recent findings from a survey on water quality.",
"The news anchor continues to stand in front of a backdrop with a map of Korea and Korean text. The text suggests the report is now on the ... | ZFWsFwpfGlE | A severe fire broke out in a Korean logistics warehouse, resulting in significant damage to the structure. An investigation into the cause of the fire and its environmental impact is ongoing. The fire has led to the release of survey results concerning water quality and waste in the affected area, highlighting the scope of the damage. Emergency responders and investigators are present, collecting evidence and conducting their duties. The investigation is in the mid-stage, with expected results to be announced. |
A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10. | [
"construction site major Korean cities",
"\"Korean construction site fires\"",
"Construction site fire at a major Korean building",
"\"Impact of fires on South Korea's infrastructure and public safety\"",
"\"South Korean authorities respond to injury in South Korea, at a time that is not specified.\"",
"\... | [
"fire construction site",
"\"construction site near me\"",
"construction site fire",
"Major fire on a specific location at a specific time",
"\"people injured\"",
"\"49 dead, 17 injured in 2023 mass shooting in [location]",
"\"Fire at [location] on [time]\"",
"\"injured person in [location] at unspeci... | [
"Construction site fires - major",
"\"construction site fire for injured in [location]\"",
"Construction site fire near [location] on [time]",
"\"major fire in any location\"",
"I couldn't find any available information that combines 'people killed' with the provided details.",
"people killed",
"\"major... | {
"article_url": "https://ko.wikipedia.org/wiki/%EC%9D%B4%EC%B2%9C_%EB%AC%BC%EB%A5%98%EC%84%BC%ED%84%B0_%EA%B3%B5%EC%82%AC%EC%9E%A5_%ED%99%94%EC%9E%AC",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION: \nExtracting events from the query that could be visually represented in a video is not possible because the query contains specific dates and locations that are difficult to visualize. This query seems to be a news event rather than a video, and it requires a specific date and location.\n\nHowever, it seems logical to extract events related to the query, as they are likely to be of interest to a video viewer.\n\nEVENTS:\n1. Construction site fire\n2. People killed\n3. Injuries\n4. South Korea",
"en_article_excerpt": "A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.",
"en_article_url": "https://en.wikipedia.org/wiki/2020_Icheon_fire",
"event_category": "disasters",
"event_name": "icheon_fire",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThe given query contains vague language about a significant event, \"a major fire,\" which could be interpreted in different ways. To extract concrete events related to the query, we need to narrow down the context. Since the event occurred in Icheon, South Korea, we will only consider events that are directly related to that location. Furthermore, given the date \"29 April 2020,\" we need to find events that are relevant to a major fire occurring in that time frame.\n\nEVENTS:\n1. Construction site fire at a major Korean building \n2. South Korean authorities respond to the fire \n3. Firefighters attempt to extinguish the blaze \n4. Rescue efforts underway for injured workers \n5. Emergency services declare the scene",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\n\nThe given query describes a tragic event involving a fire that occurred at a construction site in Icheon, South Korea. This event is highly relevant for a video search query as it involves a specific location and a sense of tragedy and loss.\n\nEVENTS:\n\n1. Construction site fire\n2. Human casualties (39 people killed)\n3. Injury to another person",
"video_URL": "http://youtube.com/watch?v=2SCMvAXvBLM",
"video_description": "경기도 이천 모가면의 한 물류창고 공사현장에서 29일 오후 1시32분 쯤 불이 나 현재까지 5명이 사망한 것으로 확인되고 있습니다. 소방당국은 대응 2단계를 발령하고 소방대원 60명과 펌프차 등 장비 27대를 투입해 진화작업을 벌이고 있습니다. 소방 관계자에 따르면 물류창고의 지하 공사현장에서 불이 시작된 것으로 보이며, 사망자는 앞으로 더 늘어날 전망입니다. 소방 대응단계는 화재 규모에 따라 1~3단계로 나뉘어지는데, 1단계는 4개 이하 소방서가 합동 대응하며 2단계는 5~9개 소방서, 3단계는 10개 이상 소방서가 함께 진화작업을 벌입니다. https://imnews.imbc.com/original/todayfield/5749059_32167.html #경기도 #이천 #물류창고 #화재 #소방당국 #대응 2단계 #진화작업",
"video_id": "2SCMvAXvBLM"
} | 16 | [
"A building is on fire with thick black smoke rising into the air. The scene shows a rural or semi-industrial area with other buildings and open land visible. The text in Korean reads \"4월 29일 경기도 이천 물류창고 화재 현장,\" indicating that the fire occurred on April 29th at a logistics warehouse in Icheon, Gyeonggi-do.",
"... | 2SCMvAXvBLM | A massive fire broke out at a logistics warehouse in Icheon, Gyeonggi-do, South Korea on April 29th. The warehouse caught fire, releasing thick black smoke into the air. The fire spread rapidly, engulfing the warehouse and surrounding buildings, and prompting emergency responders to rush to the scene. Firefighters battled the blaze, but it continued to burn fiercely. Ongoing efforts were made to extinguish the fire, with multiple fire trucks present on the scene. |
A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10. | [
"construction site major Korean cities",
"\"Korean construction site fires\"",
"Construction site fire at a major Korean building",
"\"Impact of fires on South Korea's infrastructure and public safety\"",
"\"South Korean authorities respond to injury in South Korea, at a time that is not specified.\"",
"\... | [
"fire construction site",
"\"construction site near me\"",
"construction site fire",
"Major fire on a specific location at a specific time",
"\"people injured\"",
"\"49 dead, 17 injured in 2023 mass shooting in [location]",
"\"Fire at [location] on [time]\"",
"\"injured person in [location] at unspeci... | [
"Construction site fires - major",
"\"construction site fire for injured in [location]\"",
"Construction site fire near [location] on [time]",
"\"major fire in any location\"",
"I couldn't find any available information that combines 'people killed' with the provided details.",
"people killed",
"\"major... | {
"article_url": "https://ko.wikipedia.org/wiki/%EC%9D%B4%EC%B2%9C_%EB%AC%BC%EB%A5%98%EC%84%BC%ED%84%B0_%EA%B3%B5%EC%82%AC%EC%9E%A5_%ED%99%94%EC%9E%AC",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION: \nExtracting events from the query that could be visually represented in a video is not possible because the query contains specific dates and locations that are difficult to visualize. This query seems to be a news event rather than a video, and it requires a specific date and location.\n\nHowever, it seems logical to extract events related to the query, as they are likely to be of interest to a video viewer.\n\nEVENTS:\n1. Construction site fire\n2. People killed\n3. Injuries\n4. South Korea",
"en_article_excerpt": "A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.",
"en_article_url": "https://en.wikipedia.org/wiki/2020_Icheon_fire",
"event_category": "disasters",
"event_name": "icheon_fire",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThe given query contains vague language about a significant event, \"a major fire,\" which could be interpreted in different ways. To extract concrete events related to the query, we need to narrow down the context. Since the event occurred in Icheon, South Korea, we will only consider events that are directly related to that location. Furthermore, given the date \"29 April 2020,\" we need to find events that are relevant to a major fire occurring in that time frame.\n\nEVENTS:\n1. Construction site fire at a major Korean building \n2. South Korean authorities respond to the fire \n3. Firefighters attempt to extinguish the blaze \n4. Rescue efforts underway for injured workers \n5. Emergency services declare the scene",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\n\nThe given query describes a tragic event involving a fire that occurred at a construction site in Icheon, South Korea. This event is highly relevant for a video search query as it involves a specific location and a sense of tragedy and loss.\n\nEVENTS:\n\n1. Construction site fire\n2. Human casualties (39 people killed)\n3. Injury to another person",
"video_URL": "http://youtube.com/watch?v=n8dcNr14bm8",
"video_description": "38명이 숨진 경기도 이천 물류창고 공사장 화재 현장에서 경찰이 아직 수습되지 않은 유해 일부와 유류품 등을 찾기 위한 정밀 수색을 하고 있습니다. 경기남부지방경찰청 과학수사대는 오늘(2일) 오전 9시부터 중장비와 과학수사요원 15명을 투입해 이천시 모가면 물류창고 공사 현장을 정밀 수색하고 있습니다. 오늘 수색은 사고로 숨진 사망자들의 유해 가운데 아직 수습되지 않은 일부와 유류품을 찾는 데 주력한다는 방침입니다. 앞서 경찰은 국립과학수사연구원에 의뢰한 DNA 조사를 통해 신원이 확인되지 않은 9명 가운데 8명의 신원을 확인해 유가족들에게 전했습니다. 경찰은 남은 신원 미확인자 1명의 DNA 검사 결과도 오늘 안에 나올 것으로 기대하고 있습니다. ※ '당신의 제보가 뉴스가 됩니다' YTN은 여러분의 소중한 제보를 기다립니다. [카카오톡] YTN을 검색해 채널 추가 [전화] 02-398-8585 [메일] social@ytn.co.kr [온라인 제보] www.ytn.co.kr ▶ 기사 원문 : https://www.ytn.co.kr/_ln/0103_202005021020478614 ▶ 제보 안내 : http://goo.gl/gEvsAL, 모바일앱, social@ytn.co.kr, #2424 ▣ YTN 유튜브 채널 구독 : http://goo.gl/Ytb5SZ ⓒ YTN & YTN PLUS 무단 전재 및 재배포금지",
"video_id": "n8dcNr14bm8"
} | 16 | [
"A news broadcast features two anchors in front of a large screen displaying an image of a building with visible damage, likely from a fire. The text on the screen reads \"경찰, 이번 물류창고 화재현장 정밀 수색\" (Police, conducting a detailed search at this logistics warehouse fire scene). Emergency vehicles are parked near the b... | n8dcNr14bm8 | A logistics warehouse fire broke out at a major storage facility in South Korea, and a police operation is ongoing to investigate and clean up the scene. The police are conducting a detailed search of the warehouse, and investigators are inspecting the site for signs of the fire. The investigation appears to be focused on determining the cause of the fire and identifying the responsible party. |
A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10. | [
"construction site major Korean cities",
"\"Korean construction site fires\"",
"Construction site fire at a major Korean building",
"\"Impact of fires on South Korea's infrastructure and public safety\"",
"\"South Korean authorities respond to injury in South Korea, at a time that is not specified.\"",
"\... | [
"fire construction site",
"\"construction site near me\"",
"construction site fire",
"Major fire on a specific location at a specific time",
"\"people injured\"",
"\"49 dead, 17 injured in 2023 mass shooting in [location]",
"\"Fire at [location] on [time]\"",
"\"injured person in [location] at unspeci... | [
"Construction site fires - major",
"\"construction site fire for injured in [location]\"",
"Construction site fire near [location] on [time]",
"\"major fire in any location\"",
"I couldn't find any available information that combines 'people killed' with the provided details.",
"people killed",
"\"major... | {
"article_url": "https://ko.wikipedia.org/wiki/%EC%9D%B4%EC%B2%9C_%EB%AC%BC%EB%A5%98%EC%84%BC%ED%84%B0_%EA%B3%B5%EC%82%AC%EC%9E%A5_%ED%99%94%EC%9E%AC",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION: \nExtracting events from the query that could be visually represented in a video is not possible because the query contains specific dates and locations that are difficult to visualize. This query seems to be a news event rather than a video, and it requires a specific date and location.\n\nHowever, it seems logical to extract events related to the query, as they are likely to be of interest to a video viewer.\n\nEVENTS:\n1. Construction site fire\n2. People killed\n3. Injuries\n4. South Korea",
"en_article_excerpt": "A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.",
"en_article_url": "https://en.wikipedia.org/wiki/2020_Icheon_fire",
"event_category": "disasters",
"event_name": "icheon_fire",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThe given query contains vague language about a significant event, \"a major fire,\" which could be interpreted in different ways. To extract concrete events related to the query, we need to narrow down the context. Since the event occurred in Icheon, South Korea, we will only consider events that are directly related to that location. Furthermore, given the date \"29 April 2020,\" we need to find events that are relevant to a major fire occurring in that time frame.\n\nEVENTS:\n1. Construction site fire at a major Korean building \n2. South Korean authorities respond to the fire \n3. Firefighters attempt to extinguish the blaze \n4. Rescue efforts underway for injured workers \n5. Emergency services declare the scene",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\n\nThe given query describes a tragic event involving a fire that occurred at a construction site in Icheon, South Korea. This event is highly relevant for a video search query as it involves a specific location and a sense of tragedy and loss.\n\nEVENTS:\n\n1. Construction site fire\n2. Human casualties (39 people killed)\n3. Injury to another person",
"video_URL": "http://youtube.com/watch?v=Z1TmlndGWrg",
"video_description": "대규모 인명 피해가 발생한 경기 이천 물류센터 공사장 화재와 관련해 경찰이 전담 수사본부를 편성했습니다. 경기 남부지방경찰청은 광역수사대와 과학수사대 등 125명 규모의 전담 수사본부를 편성했다고 밝혔습니다. 수사본부는 오늘(30일) 오전 10시 반 국립과학수사연구원, 소방당국과 함께 합동으로 현장감식을 진행할 예정입니다. 또, 공사 과정에서 소방법이나 건축법을 위반한 사항이 있는지 확인할 방침입니다. 신준명[shinjm7529@ytn.co.kr] ※ '당신의 제보가 뉴스가 됩니다' YTN은 여러분의 소중한 제보를 기다립니다. [카카오톡] YTN을 검색해 채널 추가 [전화] 02-398-8585 [메일] social@ytn.co.kr [온라인 제보] www.ytn.co.kr ▶ 기사 원문 : https://www.ytn.co.kr/_ln/0103_202004300027155233 ▶ 제보 안내 : http://goo.gl/gEvsAL, 모바일앱, social@ytn.co.kr, #2424 ▣ YTN 유튜브 채널 구독 : http://goo.gl/Ytb5SZ ⓒ YTN & YTN PLUS 무단 전재 및 재배포금지",
"video_id": "Z1TmlndGWrg"
} | 16 | [
"The image shows a large industrial or warehouse building with multiple white and blue panels. In front of the building, there are several ambulances and emergency vehicles parked on the road. The text in Korean at the top of the image reads \"뉴스타프 유독가스에 의식 잃은 둘\" (Newstar Staff, Two people lost consciousness due t... | Z1TmlndGWrg | The key aspects of the video depict a severe industrial accident at a logistics center, where multiple vehicles and personnel respond to an emergency situation. The incident involves toxic gas, resulting in two people losing consciousness and a fire that severely damages the building. The video shows emergency responders, including ambulances and firefighters, responding to the situation and coordinating their efforts. |
A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10. | [
"construction site major Korean cities",
"\"Korean construction site fires\"",
"Construction site fire at a major Korean building",
"\"Impact of fires on South Korea's infrastructure and public safety\"",
"\"South Korean authorities respond to injury in South Korea, at a time that is not specified.\"",
"\... | [
"fire construction site",
"\"construction site near me\"",
"construction site fire",
"Major fire on a specific location at a specific time",
"\"people injured\"",
"\"49 dead, 17 injured in 2023 mass shooting in [location]",
"\"Fire at [location] on [time]\"",
"\"injured person in [location] at unspeci... | [
"Construction site fires - major",
"\"construction site fire for injured in [location]\"",
"Construction site fire near [location] on [time]",
"\"major fire in any location\"",
"I couldn't find any available information that combines 'people killed' with the provided details.",
"people killed",
"\"major... | {
"article_url": "https://ko.wikipedia.org/wiki/%EC%9D%B4%EC%B2%9C_%EB%AC%BC%EB%A5%98%EC%84%BC%ED%84%B0_%EA%B3%B5%EC%82%AC%EC%9E%A5_%ED%99%94%EC%9E%AC",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION: \nExtracting events from the query that could be visually represented in a video is not possible because the query contains specific dates and locations that are difficult to visualize. This query seems to be a news event rather than a video, and it requires a specific date and location.\n\nHowever, it seems logical to extract events related to the query, as they are likely to be of interest to a video viewer.\n\nEVENTS:\n1. Construction site fire\n2. People killed\n3. Injuries\n4. South Korea",
"en_article_excerpt": "A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.",
"en_article_url": "https://en.wikipedia.org/wiki/2020_Icheon_fire",
"event_category": "disasters",
"event_name": "icheon_fire",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThe given query contains vague language about a significant event, \"a major fire,\" which could be interpreted in different ways. To extract concrete events related to the query, we need to narrow down the context. Since the event occurred in Icheon, South Korea, we will only consider events that are directly related to that location. Furthermore, given the date \"29 April 2020,\" we need to find events that are relevant to a major fire occurring in that time frame.\n\nEVENTS:\n1. Construction site fire at a major Korean building \n2. South Korean authorities respond to the fire \n3. Firefighters attempt to extinguish the blaze \n4. Rescue efforts underway for injured workers \n5. Emergency services declare the scene",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: A major fire occurred on 29 April 2020 at a construction site in Icheon, South Korea, killing 39 people and injuring another 10.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\n\nThe given query describes a tragic event involving a fire that occurred at a construction site in Icheon, South Korea. This event is highly relevant for a video search query as it involves a specific location and a sense of tragedy and loss.\n\nEVENTS:\n\n1. Construction site fire\n2. Human casualties (39 people killed)\n3. Injury to another person",
"video_URL": "http://youtube.com/watch?v=tHSh4jwdIJg",
"video_description": "29일 오후 1시 반쯤 경기도 이천시 모가면의 한 물류창고 공사현장에서 불이 났습니다. 이 불로 오후 5시 기준 25명이 숨진 것으로 소방당국은 추정하고 있습니다. 불이 났을 당시 현장에서 9개 업체, 80여 명이 작업 중이던 것으로 알려져 인명 피해가 더 커질 가능성도 있는 상황입니다. 물류창고 지하 2층에서 불이 시작된 것으로 추정하며, 소방당국은 \"불이 빠르게 확산되는 것으로 봐서 가연성 물질인 우레탄폼 작업 중 화재가 난 것으로 추정한다\"고 말했습니다. 현장 영상을 전해드립니다. 이천 물류창고 화재 현장 상황…소방당국 \"우레탄폼 원인 추정\" / 비디오머그 ✔신뢰할 수 있는 재미, 비디오머그 VIDEOMUG #비디오머그 #이천 #물류창고_화재 ▶비디오머그 채널 구독하기 : https://n.sbs.co.kr/mugTube ▶Homepage : https://n.sbs.co.kr/videomug ▶Facebook : https://www.facebook.com/videomug ▶Twitter : https://twitter.com/videomug_sbs",
"video_id": "tHSh4jwdIJg"
} | 16 | [
"The image shows a large building engulfed in thick black smoke, indicating a significant fire. Firefighters are visible on the ground, actively working to control the blaze. The date \"4월 29일\" (April 29th) and location \"경기 이천시 모가면\" (Icheon, Gyeonggi Province, Mogam-myeon) are indicated on the screen, suggesting... | tHSh4jwdIJg | A fire occurred at a construction site for a new warehouse in Mogam-myeon, Icheon, Gyeonggi Province, South Korea on April 29th. The fire started at approximately 1:30 AM and was extinguished by emergency responders. The fire was believed to be caused by a lightning strike. Several companies and over 80 workers were involved in the construction, and there was a possibility of additional human casualties. The fire damaged the construction site, with thick, dark smoke billowing into the sky and obscuring the surrounding area. The fire was largely contained, but the area remains inaccessible due to dense smoke and the layout of the facility. Emergency responders, including firefighters and police, worked to manage the situation and ensure public safety. |
On 9 June 2022, a fire caused by a suspected arson attack broke out at an office building in Daegu, South Korea, killing seven people and injuring 49 others. | [
"\"Fire, assassination attempt in Daegu, South Korea\"",
"\"Assassination attempted in Daegu, South Korea at a location without a specific time\"",
"\"Assassination attempt in Daegu, South Korea on [time]\"",
"\"Assassination attempt in South Korea for injured people during the [Time] in Daegu\"",
"\"suicid... | [
"\"South Korean fire festivals\"",
"\"If I'm looking for an arson attack in Daegu, South Korea at a time when it's most likely to be happening, I'd search for 'Daegu arson investigation or Daegu arson attack'\"",
"\"fire-related events in Daegu, South Korea at unspecified time\"",
"\"fire accidents in Daegu, ... | [
"\"Fire incidents in Daegu, South Korea\"",
"\"fire incidents in daegu south korea\"",
"A fire breaking out at unknown time",
"\"Fire breaking out in Daegu, South Korea during an event\"",
"arson in South Korea, Daegu, at any time",
"arson attack in Daegu, South Korea",
"\"Guilty conviction of suspiciou... | {
"article_url": "https://namu.wiki/w/%EB%8C%80%EA%B5%AC%20%EB%B3%80%ED%98%B8%EC%82%AC%20%EC%82%AC%EB%AC%B4%EC%8B%A4%20%EB%B0%A9%ED%99%94%20%EC%82%AC%EA%B1%B4",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 9 June 2022, a fire caused by a suspected arson attack broke out at an office building in Daegu, South Korea, killing seven people and injuring 49 others.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nBased on the given query, the following events could be relevant to be visualized in a video:\n \n1. A suspected arson attack at an office building.\n2. Fire breaking out.\n3. Seven people injured.\n4. 49 people killed.\n \nEVENTS:\n1. A fire breaking out\n2. A suspected arson attack\n3. Seven people injured\n4. 49 people killed",
"en_article_excerpt": "On 9 June 2022, a fire caused by a suspected arson attack broke out at an office building in Daegu, South Korea, killing seven people and injuring 49 others.",
"en_article_url": "https://en.wikipedia.org/wiki/Daegu_office_fire",
"event_category": "disasters",
"event_name": "daegu_office_fire",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 9 June 2022, a fire caused by a suspected arson attack broke out at an office building in Daegu, South Korea, killing seven people and injuring 49 others.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\n\nIt makes sense to extract the following events related to the query: \n1. Assassination or attempted assassination\n2. Suicide bombing\n3. Armed robbery\n\nThese events have clear and specific consequences that are likely to be depicted in a video, such as crime scenes, aftermath, and reactions from witnesses. They have a strong connection to the prompt's context, given the mention of a suspected arson attack and the potential for violence.\n\nEVENTS:\n\n1. Assassination attempt\n2. Suicide bombing\n3. Armed robbery",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 9 June 2022, a fire caused by a suspected arson attack broke out at an office building in Daegu, South Korea, killing seven people and injuring 49 others.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "\"EXPLANATION:\" \nAnalyzing the query, I determined that extracting events related to \"fire\" and \"death\" is a reasonable step, as these are concrete and likely to be visually represented in a video.\n\nEVENTS:\n1. Fire\n2. Death\n3. Building fire\n4. arson attack\n5. Explosion",
"video_URL": "http://youtube.com/watch?v=I2CxZCzE5pE",
"video_description": "지금부턴 어제(9일) 대구의 한 변호사 사무실에서 난 화재 관련해서 저희가 새롭게 취재한 내용 전해드리겠습니다. 저희 취재진이 사고가 난 건물의 내부 영상을 입수했습니다. 불이 시작된 건물 2층은 형체를 알아보기 어려울 정도로 검게 타버린 모습이었습니다. ▶ 기사 전문 https://news.jtbc.joins.com/html/151/NB12062151.html ▶ 뉴스룸 다시보기 (https://bit.ly/2nxI8jQ) #이승환기자 #JTBC뉴스룸 ☞JTBC뉴스레터 구독하기 (https://bit.ly/3IOFY5H) ☞JTBC유튜브 구독하기 (https://bit.ly/2hYgWZg) ☞JTBC유튜브 커뮤니티 (https://bit.ly/2LZIwke) #JTBC뉴스 공식 페이지 (홈페이지) https://news.jtbc.joins.com (APP) https://bit.ly/1r04W2D 페이스북 https://www.facebook.com/jtbcnews 트위터 https://twitter.com/JTBC_news 인스타그램 https://www.instagram.com/jtbcnews ☏ 제보하기 https://bit.ly/1krluzF 방송사 : JTBC (https://jtbc.joins.com)",
"video_id": "I2CxZCzE5pE"
} | 16 | [
"A news anchor is reporting in a studio, with a background image showing a scene of several individuals wearing bright green vests and helmets, indicating a construction or emergency site. The background image includes a green banner with Korean text that translates to \"Please do not cross!\" The lower text overla... | I2CxZCzE5pE | The video is about a significant incident involving a fire at a construction site, with multiple scenes depicting the events surrounding the fire. Key actions include:
- A fierce clash between individuals and the head of the construction company, with several individuals wearing protective gear.
- Investigation into the cause of the fire, including a damaged staircase and charred interior space.
- A severe fire that damaged an office building with significant soot and soot-covered walls and ceiling.
- Emergency responders, investigators, and officials in various uniforms, including a lawyer in the same office building.
- A meeting with the lawyer, possibly discussing the investigation and updates on the fire incident.
- The transmission to the general public, with messages about fire safety, updates, and details of the investigation.
The video is primarily informative, using various settings to describe the scene of the fire, its investigation, and the response. |
On 9 June 2022, a fire caused by a suspected arson attack broke out at an office building in Daegu, South Korea, killing seven people and injuring 49 others. | [
"\"Fire, assassination attempt in Daegu, South Korea\"",
"\"Assassination attempted in Daegu, South Korea at a location without a specific time\"",
"\"Assassination attempt in Daegu, South Korea on [time]\"",
"\"Assassination attempt in South Korea for injured people during the [Time] in Daegu\"",
"\"suicid... | [
"\"South Korean fire festivals\"",
"\"If I'm looking for an arson attack in Daegu, South Korea at a time when it's most likely to be happening, I'd search for 'Daegu arson investigation or Daegu arson attack'\"",
"\"fire-related events in Daegu, South Korea at unspecified time\"",
"\"fire accidents in Daegu, ... | [
"\"Fire incidents in Daegu, South Korea\"",
"\"fire incidents in daegu south korea\"",
"A fire breaking out at unknown time",
"\"Fire breaking out in Daegu, South Korea during an event\"",
"arson in South Korea, Daegu, at any time",
"arson attack in Daegu, South Korea",
"\"Guilty conviction of suspiciou... | {
"article_url": "https://namu.wiki/w/%EB%8C%80%EA%B5%AC%20%EB%B3%80%ED%98%B8%EC%82%AC%20%EC%82%AC%EB%AC%B4%EC%8B%A4%20%EB%B0%A9%ED%99%94%20%EC%82%AC%EA%B1%B4",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 9 June 2022, a fire caused by a suspected arson attack broke out at an office building in Daegu, South Korea, killing seven people and injuring 49 others.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nBased on the given query, the following events could be relevant to be visualized in a video:\n \n1. A suspected arson attack at an office building.\n2. Fire breaking out.\n3. Seven people injured.\n4. 49 people killed.\n \nEVENTS:\n1. A fire breaking out\n2. A suspected arson attack\n3. Seven people injured\n4. 49 people killed",
"en_article_excerpt": "On 9 June 2022, a fire caused by a suspected arson attack broke out at an office building in Daegu, South Korea, killing seven people and injuring 49 others.",
"en_article_url": "https://en.wikipedia.org/wiki/Daegu_office_fire",
"event_category": "disasters",
"event_name": "daegu_office_fire",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 9 June 2022, a fire caused by a suspected arson attack broke out at an office building in Daegu, South Korea, killing seven people and injuring 49 others.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\n\nIt makes sense to extract the following events related to the query: \n1. Assassination or attempted assassination\n2. Suicide bombing\n3. Armed robbery\n\nThese events have clear and specific consequences that are likely to be depicted in a video, such as crime scenes, aftermath, and reactions from witnesses. They have a strong connection to the prompt's context, given the mention of a suspected arson attack and the potential for violence.\n\nEVENTS:\n\n1. Assassination attempt\n2. Suicide bombing\n3. Armed robbery",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 9 June 2022, a fire caused by a suspected arson attack broke out at an office building in Daegu, South Korea, killing seven people and injuring 49 others.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "\"EXPLANATION:\" \nAnalyzing the query, I determined that extracting events related to \"fire\" and \"death\" is a reasonable step, as these are concrete and likely to be visually represented in a video.\n\nEVENTS:\n1. Fire\n2. Death\n3. Building fire\n4. arson attack\n5. Explosion",
"video_URL": "http://youtube.com/watch?v=oVzTD-Qy5HM",
"video_description": "대구 변호사 사무실 빌딩 방화 추정 화재…7명 사망 [앵커] 오늘(9일) 낮 대구 지법 근처의 한 변호사 사무실에서 방화로 보이는 화재가 발생해, 방화 용의자 1명을 포함해 7명이 숨지고 40여 명이 다쳤습니다. 현장 취재기자 연결해 자세한 소식 들어봅니다. 정지훈 기자. [기자] 네, 대구 변호사 사무실 빌딩 화재 현장에 나와 있습니다. 제 바로 뒤편 건물을 보시면 전면 유리창들이 깨진 걸 보실 수 있을텐데요. 신고를 받고 출동한 소방당국이 화재 진압과 구조를 위해 일부러 깨드린 겁니다. 당시 상황이 얼마나 긴박했는지를 알 수 있습니다. 이 건물은 지하 2층, 지상 5층짜리 건물로 오늘(9일) 오전 10시 55분쯤 건물 2층에서 불이 났습니다. 불은 소방 당국에 의해 20여 분만에 꺼졌지만, 남자 5명, 여자 2명 등 7명이 숨지고 40여 명이 화상 또는 연기를 흡입해 병원 치료를 받고 있습니다. 이들 사망자는 2층 203호에서 모두 발견됐는데요. 화재 발생 당시 119에는 \"건물 2층에서 검은 연기가 보이고 폭발음이 들렸다\"는 신고가 접수됐습니다. 사망자의 신원이 아직 확인되지 않았습니다. 같은 장소에서 방화 용의자도 숨진 채 발견됐습니다. 불이 나자 건물 3층에 있던 사람들은 4층 등 건물 위층으로, 일부는 건물 밖에 설치된 비상계단을 통해 옥상으로 대피하기도 했습니다. 화재 현장에 있었던 목격자의 말 들어보시겠습니다. [현장 목격자] \"어느 순간 폭발음이 팡 터지면서 유리창 깨지는 소리가 계속 나고 제가 너무 놀라서 상담실에서 뛰쳐 나와서 문을 열려고 하니 이미 뜨거웠어요. 그래서 몸으로 문을 밀쳐서 열고…\" 소방당국은 화재 진압 과정에서 급격히 연소가 확대돼 인명 피해 규모가 커졌다고 설명했습니다. 당국은 브리핑에서 \"화재 건물 지하층에만 스프링클러가 있고 지상층에는 스프링클러가 설치돼 있지 않다\"고 밝혔습니다. 화재가 발생한 빌딩은 법원 뒤쪽에 위치해 변호사 사무실이 밀집한 곳입니다. 경찰은 50대 방화 용의자가 CCTV상 주거지에서 뭔가를 들고 나오는 장면을 확인하고 자세한 화재 경위를 조사하고 있는데요. 경찰은 불이 난 변호사 사무실에 의뢰된 송사 사건의 상대인 용의자가 사무실에 불을 낸 것으로 보고, 수사전담팀을 편성해 사건 발생 경위 등을 집중 조사할 방침입니다. 지금까지 대구에서 연합뉴스TV 정지훈입니다. #대구_변호사_사무실 #폭발 #방화 연합뉴스TV 기사문의 및 제보 : 카톡/라인 jebo23 ▣ 연합뉴스TV 유튜브 채널 구독 https://goo.gl/VuCJMi ▣ 대한민국 뉴스의 시작 연합뉴스TV / Yonhap News TV http://www.yonhapnewstv.co.kr/",
"video_id": "oVzTD-Qy5HM"
} | 16 | [
"The image shows two news anchors sitting side by side in front of a world map background. They are presenting a news segment. A red banner at the bottom of the screen displays Korean text, which likely summarizes the main news headline. The time on the screen indicates it is 16:48. The text on the banner suggests ... | oVzTD-Qy5HM | A government official has been dismissed from their position, involving seven individuals in total. A series of events unfolded, starting with a news anchor and a female anchor discussing a story about the dismissal on a world map background. The dismissal involved financial information and a stock market figure, with additional details provided in the red banner. This is followed by live footage of a news segment, including a live reporter outside a building with several people and security personnel, where the dismissal was confirmed.
The scene shifted to an aerial view of the construction site where a fire incident occurred, resulting in the death of seven people. Emergency responders and officials attended to the scene, and the news continued to provide updates, including the involvement of the police in investigating the arson. The incident occurred at a building with a suspicious fire, with multiple casualties. The investigation is ongoing, with the police currently investigating the possibility of arson. |
On 9 June 2022, a fire caused by a suspected arson attack broke out at an office building in Daegu, South Korea, killing seven people and injuring 49 others. | [
"\"Fire, assassination attempt in Daegu, South Korea\"",
"\"Assassination attempted in Daegu, South Korea at a location without a specific time\"",
"\"Assassination attempt in Daegu, South Korea on [time]\"",
"\"Assassination attempt in South Korea for injured people during the [Time] in Daegu\"",
"\"suicid... | [
"\"South Korean fire festivals\"",
"\"If I'm looking for an arson attack in Daegu, South Korea at a time when it's most likely to be happening, I'd search for 'Daegu arson investigation or Daegu arson attack'\"",
"\"fire-related events in Daegu, South Korea at unspecified time\"",
"\"fire accidents in Daegu, ... | [
"\"Fire incidents in Daegu, South Korea\"",
"\"fire incidents in daegu south korea\"",
"A fire breaking out at unknown time",
"\"Fire breaking out in Daegu, South Korea during an event\"",
"arson in South Korea, Daegu, at any time",
"arson attack in Daegu, South Korea",
"\"Guilty conviction of suspiciou... | {
"article_url": "https://namu.wiki/w/%EB%8C%80%EA%B5%AC%20%EB%B3%80%ED%98%B8%EC%82%AC%20%EC%82%AC%EB%AC%B4%EC%8B%A4%20%EB%B0%A9%ED%99%94%20%EC%82%AC%EA%B1%B4",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 9 June 2022, a fire caused by a suspected arson attack broke out at an office building in Daegu, South Korea, killing seven people and injuring 49 others.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nBased on the given query, the following events could be relevant to be visualized in a video:\n \n1. A suspected arson attack at an office building.\n2. Fire breaking out.\n3. Seven people injured.\n4. 49 people killed.\n \nEVENTS:\n1. A fire breaking out\n2. A suspected arson attack\n3. Seven people injured\n4. 49 people killed",
"en_article_excerpt": "On 9 June 2022, a fire caused by a suspected arson attack broke out at an office building in Daegu, South Korea, killing seven people and injuring 49 others.",
"en_article_url": "https://en.wikipedia.org/wiki/Daegu_office_fire",
"event_category": "disasters",
"event_name": "daegu_office_fire",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 9 June 2022, a fire caused by a suspected arson attack broke out at an office building in Daegu, South Korea, killing seven people and injuring 49 others.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\n\nIt makes sense to extract the following events related to the query: \n1. Assassination or attempted assassination\n2. Suicide bombing\n3. Armed robbery\n\nThese events have clear and specific consequences that are likely to be depicted in a video, such as crime scenes, aftermath, and reactions from witnesses. They have a strong connection to the prompt's context, given the mention of a suspected arson attack and the potential for violence.\n\nEVENTS:\n\n1. Assassination attempt\n2. Suicide bombing\n3. Armed robbery",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 9 June 2022, a fire caused by a suspected arson attack broke out at an office building in Daegu, South Korea, killing seven people and injuring 49 others.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "\"EXPLANATION:\" \nAnalyzing the query, I determined that extracting events related to \"fire\" and \"death\" is a reasonable step, as these are concrete and likely to be visually represented in a video.\n\nEVENTS:\n1. Fire\n2. Death\n3. Building fire\n4. arson attack\n5. Explosion",
"video_URL": "http://twitter.com/i/status/1534839444830044161",
"video_description": "대구 변호사 사무실 화재",
"video_id": "1534839444830044161"
} | 16 | [
"The image shows a news reporter in front of a building labeled \"우정 배달 당,\" which translates to \"Wujeong Delivery Office.\" The reporter is wearing a face mask and holding a microphone with the KBS logo. In the background, a red truck and a person in a blue vest are visible. The news ticker at the bottom reads, \... | 1534839444830044161 | A building in Daegu, South Korea is the focal point of a fire that has caused significant damage, resulting in the loss of at least 7 lives. The incident occurred at the Wujeong Delivery Office, a commercial building that was damaged in the fire.
The fire, suspected to have started in the office, spread rapidly due to the combustible materials present. The rescue efforts began immediately, with emergency services rushing to the scene to assess the situation and evacuate the building. The fire is currently out of control, with multiple buildings in the area affected.
Emergency responders are working to address the aftermath of the incident, including the evacuation of occupants and damage assessments. The investigation is ongoing, focusing on the causes of the fire, the individuals involved, and the circumstances surrounding the incident.
The presence of high-visibility gear, including helmets and ladders, has raised concerns about safety and security in the building, suggesting a broader context of these issues within the building. An investigation into the incident is ongoing, with concerns about the missing security cameras and emergency lighting systems further highlighting the urgency of the situation. |
On 9 June 2022, a fire caused by a suspected arson attack broke out at an office building in Daegu, South Korea, killing seven people and injuring 49 others. | [
"\"Fire, assassination attempt in Daegu, South Korea\"",
"\"Assassination attempted in Daegu, South Korea at a location without a specific time\"",
"\"Assassination attempt in Daegu, South Korea on [time]\"",
"\"Assassination attempt in South Korea for injured people during the [Time] in Daegu\"",
"\"suicid... | [
"\"South Korean fire festivals\"",
"\"If I'm looking for an arson attack in Daegu, South Korea at a time when it's most likely to be happening, I'd search for 'Daegu arson investigation or Daegu arson attack'\"",
"\"fire-related events in Daegu, South Korea at unspecified time\"",
"\"fire accidents in Daegu, ... | [
"\"Fire incidents in Daegu, South Korea\"",
"\"fire incidents in daegu south korea\"",
"A fire breaking out at unknown time",
"\"Fire breaking out in Daegu, South Korea during an event\"",
"arson in South Korea, Daegu, at any time",
"arson attack in Daegu, South Korea",
"\"Guilty conviction of suspiciou... | {
"article_url": "https://namu.wiki/w/%EB%8C%80%EA%B5%AC%20%EB%B3%80%ED%98%B8%EC%82%AC%20%EC%82%AC%EB%AC%B4%EC%8B%A4%20%EB%B0%A9%ED%99%94%20%EC%82%AC%EA%B1%B4",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 9 June 2022, a fire caused by a suspected arson attack broke out at an office building in Daegu, South Korea, killing seven people and injuring 49 others.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nBased on the given query, the following events could be relevant to be visualized in a video:\n \n1. A suspected arson attack at an office building.\n2. Fire breaking out.\n3. Seven people injured.\n4. 49 people killed.\n \nEVENTS:\n1. A fire breaking out\n2. A suspected arson attack\n3. Seven people injured\n4. 49 people killed",
"en_article_excerpt": "On 9 June 2022, a fire caused by a suspected arson attack broke out at an office building in Daegu, South Korea, killing seven people and injuring 49 others.",
"en_article_url": "https://en.wikipedia.org/wiki/Daegu_office_fire",
"event_category": "disasters",
"event_name": "daegu_office_fire",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 9 June 2022, a fire caused by a suspected arson attack broke out at an office building in Daegu, South Korea, killing seven people and injuring 49 others.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\n\nIt makes sense to extract the following events related to the query: \n1. Assassination or attempted assassination\n2. Suicide bombing\n3. Armed robbery\n\nThese events have clear and specific consequences that are likely to be depicted in a video, such as crime scenes, aftermath, and reactions from witnesses. They have a strong connection to the prompt's context, given the mention of a suspected arson attack and the potential for violence.\n\nEVENTS:\n\n1. Assassination attempt\n2. Suicide bombing\n3. Armed robbery",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 9 June 2022, a fire caused by a suspected arson attack broke out at an office building in Daegu, South Korea, killing seven people and injuring 49 others.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "\"EXPLANATION:\" \nAnalyzing the query, I determined that extracting events related to \"fire\" and \"death\" is a reasonable step, as these are concrete and likely to be visually represented in a video.\n\nEVENTS:\n1. Fire\n2. Death\n3. Building fire\n4. arson attack\n5. Explosion",
"video_URL": "http://twitter.com/i/status/1534851387024359424",
"video_description": "흰 천 덮은 물건 든 방화 용의자…CCTV로 본 화재 순간 (출처 : JTBC) http://naver.me/xtqfdWvA 대구 방화 용의자 현장에서 숨져...\"재판 결과에 \\’앙심\\’” / YTN https://youtu.be/Msh294y6bO8 재개발투자 8억 반환소송 패소…2심 앞두고 범행",
"video_id": "1534851387024359424"
} | 16 | [
"In the image, two news anchors are seated at a desk, presenting a news segment. Behind them, there is a backdrop of an emergency scene with a red fire truck marked with the number \"4012\" and a license plate \"71가 3534.\" Several emergency personnel are visible, suggesting a response to an incident. The text in t... | 1534851387024359424 | An arson incident occurred at a Korean news network's (YTN) headquarters, resulting in the death of a suspect, triggering a major investigation. The incident began with an emergency scene at the headquarters, where a fire was reported. Emergency personnel responded and assisted in extinguishing the fire, but the suspect died in the process. The investigation is ongoing, with police forming a special investigation team. The police collected CCTV footage from surrounding areas, including a building where the suspect was seen. Rescuers and investigators were seen responding to the incident, and a red fire truck was visible at the scene. |
On 9 June 2022, a fire caused by a suspected arson attack broke out at an office building in Daegu, South Korea, killing seven people and injuring 49 others. | [
"\"Fire, assassination attempt in Daegu, South Korea\"",
"\"Assassination attempted in Daegu, South Korea at a location without a specific time\"",
"\"Assassination attempt in Daegu, South Korea on [time]\"",
"\"Assassination attempt in South Korea for injured people during the [Time] in Daegu\"",
"\"suicid... | [
"\"South Korean fire festivals\"",
"\"If I'm looking for an arson attack in Daegu, South Korea at a time when it's most likely to be happening, I'd search for 'Daegu arson investigation or Daegu arson attack'\"",
"\"fire-related events in Daegu, South Korea at unspecified time\"",
"\"fire accidents in Daegu, ... | [
"\"Fire incidents in Daegu, South Korea\"",
"\"fire incidents in daegu south korea\"",
"A fire breaking out at unknown time",
"\"Fire breaking out in Daegu, South Korea during an event\"",
"arson in South Korea, Daegu, at any time",
"arson attack in Daegu, South Korea",
"\"Guilty conviction of suspiciou... | {
"article_url": "https://namu.wiki/w/%EB%8C%80%EA%B5%AC%20%EB%B3%80%ED%98%B8%EC%82%AC%20%EC%82%AC%EB%AC%B4%EC%8B%A4%20%EB%B0%A9%ED%99%94%20%EC%82%AC%EA%B1%B4",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 9 June 2022, a fire caused by a suspected arson attack broke out at an office building in Daegu, South Korea, killing seven people and injuring 49 others.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nBased on the given query, the following events could be relevant to be visualized in a video:\n \n1. A suspected arson attack at an office building.\n2. Fire breaking out.\n3. Seven people injured.\n4. 49 people killed.\n \nEVENTS:\n1. A fire breaking out\n2. A suspected arson attack\n3. Seven people injured\n4. 49 people killed",
"en_article_excerpt": "On 9 June 2022, a fire caused by a suspected arson attack broke out at an office building in Daegu, South Korea, killing seven people and injuring 49 others.",
"en_article_url": "https://en.wikipedia.org/wiki/Daegu_office_fire",
"event_category": "disasters",
"event_name": "daegu_office_fire",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 9 June 2022, a fire caused by a suspected arson attack broke out at an office building in Daegu, South Korea, killing seven people and injuring 49 others.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\n\nIt makes sense to extract the following events related to the query: \n1. Assassination or attempted assassination\n2. Suicide bombing\n3. Armed robbery\n\nThese events have clear and specific consequences that are likely to be depicted in a video, such as crime scenes, aftermath, and reactions from witnesses. They have a strong connection to the prompt's context, given the mention of a suspected arson attack and the potential for violence.\n\nEVENTS:\n\n1. Assassination attempt\n2. Suicide bombing\n3. Armed robbery",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 9 June 2022, a fire caused by a suspected arson attack broke out at an office building in Daegu, South Korea, killing seven people and injuring 49 others.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "\"EXPLANATION:\" \nAnalyzing the query, I determined that extracting events related to \"fire\" and \"death\" is a reasonable step, as these are concrete and likely to be visually represented in a video.\n\nEVENTS:\n1. Fire\n2. Death\n3. Building fire\n4. arson attack\n5. Explosion",
"video_URL": "http://youtube.com/watch?v=rzyM4QtZNNU",
"video_description": "오늘(9일) 대구광역시 수성구 범어동 대구지방법원 인근 변호사 사무실 빌딩에서 방화로 추정되는 불이 나 7명이 숨지고 49명이 다쳤습니다. 건물 CCTV를 보면 오전 10시 53분 마스크를 쓰고 빌딩에 들어서는 남성이 확인됩니다. 한 손에는 흰 천으로 덮은 확인되지 않은 물체를 든 상태로, 경찰은 천에 덮인 물체가 인화 물질이었을 것으로 보고 있습니다. 이 남성이 해당 변호사 사무실에 들어간 지 23초 만에 불길이 일어나는 모습이 확인됩니다. ▣ KBS 기사 원문보기 : http://news.kbs.co.kr/news/view.do?ncd=5482165 ▣ 제보 하기 ◇ 카카오톡 : 'KBS제보' 검색 ◇ 전화 : 02-781-1234 ◇ 홈페이지 : https://goo.gl/4bWbkG ◇ 이메일 : kbs1234@kbs.co.kr #대구화재 #방화용의자 #CCTV",
"video_id": "rzyM4QtZNNU"
} | 16 | [
"The image shows a person wearing a blue jacket and carrying a backpack, walking down a staircase in a well-lit hallway. There are couches on the right side of the hallway, and a green arrow pointing downwards is visible in the foreground. The text in Korean reads: \"현 천으로 둔 물건 들고 건물 들어가는 방화 용의자\" which translates ... | rzyM4QtZNNU | A person carrying a backpack walks into a building, causing a suspicious person to enter the building with hidden items, and the building's hallway is well-lit. The person on the second floor then climbs the stairs and enters a building, leading to a news broadcast segment that includes a graphic with the text "KBS NEWS". The image of a smoky hallway indicates a dramatic event has occurred, with the person fleeing from the scene as smoke and flames fill the area. The scene transitions to a news broadcast, delivering updates and information related to the incident. |
On 9 June 2022, a fire caused by a suspected arson attack broke out at an office building in Daegu, South Korea, killing seven people and injuring 49 others. | [
"\"Fire, assassination attempt in Daegu, South Korea\"",
"\"Assassination attempted in Daegu, South Korea at a location without a specific time\"",
"\"Assassination attempt in Daegu, South Korea on [time]\"",
"\"Assassination attempt in South Korea for injured people during the [Time] in Daegu\"",
"\"suicid... | [
"\"South Korean fire festivals\"",
"\"If I'm looking for an arson attack in Daegu, South Korea at a time when it's most likely to be happening, I'd search for 'Daegu arson investigation or Daegu arson attack'\"",
"\"fire-related events in Daegu, South Korea at unspecified time\"",
"\"fire accidents in Daegu, ... | [
"\"Fire incidents in Daegu, South Korea\"",
"\"fire incidents in daegu south korea\"",
"A fire breaking out at unknown time",
"\"Fire breaking out in Daegu, South Korea during an event\"",
"arson in South Korea, Daegu, at any time",
"arson attack in Daegu, South Korea",
"\"Guilty conviction of suspiciou... | {
"article_url": "https://namu.wiki/w/%EB%8C%80%EA%B5%AC%20%EB%B3%80%ED%98%B8%EC%82%AC%20%EC%82%AC%EB%AC%B4%EC%8B%A4%20%EB%B0%A9%ED%99%94%20%EC%82%AC%EA%B1%B4",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 9 June 2022, a fire caused by a suspected arson attack broke out at an office building in Daegu, South Korea, killing seven people and injuring 49 others.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nBased on the given query, the following events could be relevant to be visualized in a video:\n \n1. A suspected arson attack at an office building.\n2. Fire breaking out.\n3. Seven people injured.\n4. 49 people killed.\n \nEVENTS:\n1. A fire breaking out\n2. A suspected arson attack\n3. Seven people injured\n4. 49 people killed",
"en_article_excerpt": "On 9 June 2022, a fire caused by a suspected arson attack broke out at an office building in Daegu, South Korea, killing seven people and injuring 49 others.",
"en_article_url": "https://en.wikipedia.org/wiki/Daegu_office_fire",
"event_category": "disasters",
"event_name": "daegu_office_fire",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 9 June 2022, a fire caused by a suspected arson attack broke out at an office building in Daegu, South Korea, killing seven people and injuring 49 others.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\n\nIt makes sense to extract the following events related to the query: \n1. Assassination or attempted assassination\n2. Suicide bombing\n3. Armed robbery\n\nThese events have clear and specific consequences that are likely to be depicted in a video, such as crime scenes, aftermath, and reactions from witnesses. They have a strong connection to the prompt's context, given the mention of a suspected arson attack and the potential for violence.\n\nEVENTS:\n\n1. Assassination attempt\n2. Suicide bombing\n3. Armed robbery",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: On 9 June 2022, a fire caused by a suspected arson attack broke out at an office building in Daegu, South Korea, killing seven people and injuring 49 others.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "\"EXPLANATION:\" \nAnalyzing the query, I determined that extracting events related to \"fire\" and \"death\" is a reasonable step, as these are concrete and likely to be visually represented in a video.\n\nEVENTS:\n1. Fire\n2. Death\n3. Building fire\n4. arson attack\n5. Explosion",
"video_URL": "http://youtube.com/watch?v=5xiV1Z2MDX4",
"video_description": "대구 변호사 사무실 빌딩 화재…7명 사망·46명 부상 오늘(9일) 오전 11시쯤 대구시 수성구 범어동 대구지방법원 인근 7층짜리 변호사 사무실 빌딩 2층에서 불이 나 7명이 숨지고 46명이 연기를 마시는 등 부상을 입었습니다. 또 안에 있던 수십 명이 긴급 대피했습니다. 사망자는 남성 5명, 여성 2명인 것으로 알려졌습니다. 당시 119에는 \"2층에서 검은 연기가 보이고 폭발음이 들렸다\"는 신고가 접수됐습니다. 불이 나자 소방차 50대와 소방인력 160명이 동원돼 불을 끄고 입주자들을 구조했습니다. 불은 20여 분만에 진화됐습니다. 경찰은 화재 발생 당시 의뢰인이 불만을 제기한 정황이 있었다는 점 등으로 미뤄 방화 여부를 조사하고 있습니다. #사무실_빌딩_화재 #사망 #연기 #대구_범어동 연합뉴스TV 기사문의 및 제보 : 카톡/라인 jebo23 ▣ 연합뉴스TV 유튜브 채널 구독 https://goo.gl/VuCJMi ▣ 대한민국 뉴스의 시작 연합뉴스TV / Yonhap News TV http://www.yonhapnewstv.co.kr/",
"video_id": "5xiV1Z2MDX4"
} | 16 | [
"In the current frame of the video, two news anchors are seated in a news studio, presenting the news. The male anchor on the left is wearing a suit with a pink tie, while the female anchor on the right is dressed in a light blue blazer over a white top. The backdrop features multiple screens displaying various new... | 5xiV1Z2MDX4 | A devastating fire incident occurred at a lawyer's office building in Daegu, South Korea, resulting in seven deaths and forty-six injuries. The incident is being investigated, with emergency responders, including firefighters and police officers, conducting a thorough assessment of the situation. The rescue and response efforts continue, with multiple vehicles, including a white ambulance and a fire truck, positioned near the building. The incident is being reported from a news studio, where the announcer provides updates and context about the fire and its aftermath. |
Typhoon Hinnamnor, known in the Philippines as Super Typhoon Henry, was a very large and powerful tropical cyclone that impacted Japan, South Korea, Taiwan, the Philippines, and Russia. | [
"\"Typhoon Hinnamnor making landfall in Japan around [insert time]\"",
"\"Typhoon Hinnamnor in Japan at time NOT AVAILABLE\"",
"\"Typhoon Hinnamnor in South Korea\"",
"\"Typhoon Hinnamnor making landfall in Taiwan in [time] in Japan\"",
"typhoon hinnamnor weather forecast philippines",
"\"Hinamnor Typhoon... | [
"\"Superteon\" \"Hinnamnor\"",
"typhoon hinnamnor japan at 14:30 this weekend",
"weather forecast north korea typhoon",
"\"Typhoon Hinnamnor in Taiwan 2024\"",
"typhoon hinnamnor typhoon hinnamnor philippines 1989",
"\"Typhoon Hinnamnor in Russia at unknown time\"",
"\"Super Typhoon Hinnamnor Superteon ... | [
"Typhoon Hinnamnor/Superteon Henry",
"Japan Landfall",
"\"South Korea landfall during [time]\"",
"\"Taiwan\" landfall",
"landfall philippines",
"\"Russia landfall [time]\"",
"Typhoon Hinnamnor/Superteon Henry Widespread destruction",
"Japan: Wildfires",
"\"South Korea flood 2017\"",
"\"Taiwan natu... | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%83%9C%ED%92%8D_%ED%9E%8C%EB%82%A8%EB%85%B8",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Hinnamnor, known in the Philippines as Super Typhoon Henry, was a very large and powerful tropical cyclone that impacted Japan, South Korea, Taiwan, the Philippines, and Russia.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nIt is likely to extract the following events from the given query:\n \nThese events seem to be related to the topic of Typhoon Hinnamnor in its impact on other countries:\n1. Typhoon Hinnamnor makes landfall in Japan, South Korea, Taiwan, the Philippines, and Russia.\n2. The storm causes widespread destruction, landslides and flooding in the Philippines.\n3. The storm brings heavy rainfall and strong winds that cause massive power outages in the region.\n4. The international community issues warnings about the potential for widespread damage and loss of life.\n\nEVENTS:\n1. Landfall\n2. Widespread destruction\n3. Landslides\n4. Flooding\n5. Power outages",
"en_article_excerpt": "Typhoon Hinnamnor, known in the Philippines as Super Typhoon Henry, was a very large and powerful tropical cyclone that impacted Japan, South Korea, Taiwan, the Philippines, and Russia.",
"en_article_url": "https://en.wikipedia.org/wiki/Typhoon_Hinnamnor",
"event_category": "disasters",
"event_name": "typhoon_hinnamno",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Hinnamnor, known in the Philippines as Super Typhoon Henry, was a very large and powerful tropical cyclone that impacted Japan, South Korea, Taiwan, the Philippines, and Russia.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nGiven the query \"Typhoon Hinnamnor, known as Super Typhoon Henry, was a very large and powerful tropical cyclone that impacted Japan, South Korea, Taiwan, the Philippines, and Russia\", it is reasonable to extract events that involve the typhoon impacting these countries. The query specifically mentions several countries and regions where the typhoon had significant impacts, suggesting that the events should focus on these locations.\n\nEVENTS:\n1. Typhoon Hinnamnor making landfall in Japan\n2. Typhoon Hinnamnor affecting Taiwan\n3. Typhoon Hinnamnor affecting South Korea\n4. Typhoon Hinnamnor making landfall in the Philippines\n5. Typhoon Hinnamnor causing significant damage in Russia",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Hinnamnor, known in the Philippines as Super Typhoon Henry, was a very large and powerful tropical cyclone that impacted Japan, South Korea, Taiwan, the Philippines, and Russia.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nConsidering the query \"Typhoon Hinnamnor, known in the Philippines as Super Typhoon Henry\", I would extract events that are likely to be seen in a video related to this event. These events should be concrete enough to be visualized in a video, as the query is describing a natural disaster that can be observed in various locations. Based on this, the following events seem reasonable to extract:\n\n1. \"Typhoon Hinnamnor\" \n2. \"Super Typhoon Henry\" \n3. \"Impact on Japan\" \n4. \"Impact on Korea\" \n5. \"Typhoon Hinnamnor makes landfall\" \n\nEVENTS: \n1. Typhoon Hinnamnor\n2. Super Typhoon Henry\n3. Impact on Japan\n4. Impact on Korea\n5. Typhoon Hinnamnor makes landfall",
"video_URL": "http://twitter.com/i/status/1566887846916792320",
"video_description": "이게맞나.. #태풍 #힌남노 #포항 #역대급태풍",
"video_id": "1566887846916792320"
} | 16 | [
"The image shows a nighttime scene with heavy rainfall. Several trees and a street are visible, illuminated by streetlights. The rain appears to be pouring heavily, creating a misty atmosphere. Power lines and a building are also visible in the background.",
"The current frame continues to depict a nighttime scen... | 1566887846916792320 | The video depicts a nighttime scene with heavy rainfall, highlighting a misty atmosphere and a focus on the road and surrounding trees. Key participants include vehicles, streetlights, and power lines. The overall context is an ongoing weather event, with the scene showing the progression and persistence of the rain, with the misty atmosphere maintaining its presence throughout. |
Typhoon Hinnamnor, known in the Philippines as Super Typhoon Henry, was a very large and powerful tropical cyclone that impacted Japan, South Korea, Taiwan, the Philippines, and Russia. | [
"\"Typhoon Hinnamnor making landfall in Japan around [insert time]\"",
"\"Typhoon Hinnamnor in Japan at time NOT AVAILABLE\"",
"\"Typhoon Hinnamnor in South Korea\"",
"\"Typhoon Hinnamnor making landfall in Taiwan in [time] in Japan\"",
"typhoon hinnamnor weather forecast philippines",
"\"Hinamnor Typhoon... | [
"\"Superteon\" \"Hinnamnor\"",
"typhoon hinnamnor japan at 14:30 this weekend",
"weather forecast north korea typhoon",
"\"Typhoon Hinnamnor in Taiwan 2024\"",
"typhoon hinnamnor typhoon hinnamnor philippines 1989",
"\"Typhoon Hinnamnor in Russia at unknown time\"",
"\"Super Typhoon Hinnamnor Superteon ... | [
"Typhoon Hinnamnor/Superteon Henry",
"Japan Landfall",
"\"South Korea landfall during [time]\"",
"\"Taiwan\" landfall",
"landfall philippines",
"\"Russia landfall [time]\"",
"Typhoon Hinnamnor/Superteon Henry Widespread destruction",
"Japan: Wildfires",
"\"South Korea flood 2017\"",
"\"Taiwan natu... | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%83%9C%ED%92%8D_%ED%9E%8C%EB%82%A8%EB%85%B8",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Hinnamnor, known in the Philippines as Super Typhoon Henry, was a very large and powerful tropical cyclone that impacted Japan, South Korea, Taiwan, the Philippines, and Russia.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nIt is likely to extract the following events from the given query:\n \nThese events seem to be related to the topic of Typhoon Hinnamnor in its impact on other countries:\n1. Typhoon Hinnamnor makes landfall in Japan, South Korea, Taiwan, the Philippines, and Russia.\n2. The storm causes widespread destruction, landslides and flooding in the Philippines.\n3. The storm brings heavy rainfall and strong winds that cause massive power outages in the region.\n4. The international community issues warnings about the potential for widespread damage and loss of life.\n\nEVENTS:\n1. Landfall\n2. Widespread destruction\n3. Landslides\n4. Flooding\n5. Power outages",
"en_article_excerpt": "Typhoon Hinnamnor, known in the Philippines as Super Typhoon Henry, was a very large and powerful tropical cyclone that impacted Japan, South Korea, Taiwan, the Philippines, and Russia.",
"en_article_url": "https://en.wikipedia.org/wiki/Typhoon_Hinnamnor",
"event_category": "disasters",
"event_name": "typhoon_hinnamno",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Hinnamnor, known in the Philippines as Super Typhoon Henry, was a very large and powerful tropical cyclone that impacted Japan, South Korea, Taiwan, the Philippines, and Russia.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nGiven the query \"Typhoon Hinnamnor, known as Super Typhoon Henry, was a very large and powerful tropical cyclone that impacted Japan, South Korea, Taiwan, the Philippines, and Russia\", it is reasonable to extract events that involve the typhoon impacting these countries. The query specifically mentions several countries and regions where the typhoon had significant impacts, suggesting that the events should focus on these locations.\n\nEVENTS:\n1. Typhoon Hinnamnor making landfall in Japan\n2. Typhoon Hinnamnor affecting Taiwan\n3. Typhoon Hinnamnor affecting South Korea\n4. Typhoon Hinnamnor making landfall in the Philippines\n5. Typhoon Hinnamnor causing significant damage in Russia",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Hinnamnor, known in the Philippines as Super Typhoon Henry, was a very large and powerful tropical cyclone that impacted Japan, South Korea, Taiwan, the Philippines, and Russia.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nConsidering the query \"Typhoon Hinnamnor, known in the Philippines as Super Typhoon Henry\", I would extract events that are likely to be seen in a video related to this event. These events should be concrete enough to be visualized in a video, as the query is describing a natural disaster that can be observed in various locations. Based on this, the following events seem reasonable to extract:\n\n1. \"Typhoon Hinnamnor\" \n2. \"Super Typhoon Henry\" \n3. \"Impact on Japan\" \n4. \"Impact on Korea\" \n5. \"Typhoon Hinnamnor makes landfall\" \n\nEVENTS: \n1. Typhoon Hinnamnor\n2. Super Typhoon Henry\n3. Impact on Japan\n4. Impact on Korea\n5. Typhoon Hinnamnor makes landfall",
"video_URL": "http://twitter.com/i/status/1566774984525025281",
"video_description": "다들... 이번 태풍 힌남노 조심하세요... 현재 제주... 저희 집 나무가 90도로 순간적으로 휘어진 거 처음 봐요 바람 장난아니고 비도 장난아니니까 집에... 콕... 박혀있기",
"video_id": "1566774984525025281"
} | 16 | [
"The image is too dark to discern specific details. There are some silhouettes of trees and possibly a building in the background, but no clear actions, participants, or context can be identified due to low visibility.",
"The current frame remains very dark, with low visibility making it difficult to discern spec... | 1566774984525025281 | The video depicts a scene of a deserted, dimly lit forest with a building in the distance. The scene lacks clear details, and only silhouettes of trees and the building are visible. The lack of visibility and distinct actions or participants persists throughout the video, creating a sense of unease and mystery. |
Typhoon Hinnamnor, known in the Philippines as Super Typhoon Henry, was a very large and powerful tropical cyclone that impacted Japan, South Korea, Taiwan, the Philippines, and Russia. | [
"\"Typhoon Hinnamnor making landfall in Japan around [insert time]\"",
"\"Typhoon Hinnamnor in Japan at time NOT AVAILABLE\"",
"\"Typhoon Hinnamnor in South Korea\"",
"\"Typhoon Hinnamnor making landfall in Taiwan in [time] in Japan\"",
"typhoon hinnamnor weather forecast philippines",
"\"Hinamnor Typhoon... | [
"\"Superteon\" \"Hinnamnor\"",
"typhoon hinnamnor japan at 14:30 this weekend",
"weather forecast north korea typhoon",
"\"Typhoon Hinnamnor in Taiwan 2024\"",
"typhoon hinnamnor typhoon hinnamnor philippines 1989",
"\"Typhoon Hinnamnor in Russia at unknown time\"",
"\"Super Typhoon Hinnamnor Superteon ... | [
"Typhoon Hinnamnor/Superteon Henry",
"Japan Landfall",
"\"South Korea landfall during [time]\"",
"\"Taiwan\" landfall",
"landfall philippines",
"\"Russia landfall [time]\"",
"Typhoon Hinnamnor/Superteon Henry Widespread destruction",
"Japan: Wildfires",
"\"South Korea flood 2017\"",
"\"Taiwan natu... | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%83%9C%ED%92%8D_%ED%9E%8C%EB%82%A8%EB%85%B8",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Hinnamnor, known in the Philippines as Super Typhoon Henry, was a very large and powerful tropical cyclone that impacted Japan, South Korea, Taiwan, the Philippines, and Russia.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nIt is likely to extract the following events from the given query:\n \nThese events seem to be related to the topic of Typhoon Hinnamnor in its impact on other countries:\n1. Typhoon Hinnamnor makes landfall in Japan, South Korea, Taiwan, the Philippines, and Russia.\n2. The storm causes widespread destruction, landslides and flooding in the Philippines.\n3. The storm brings heavy rainfall and strong winds that cause massive power outages in the region.\n4. The international community issues warnings about the potential for widespread damage and loss of life.\n\nEVENTS:\n1. Landfall\n2. Widespread destruction\n3. Landslides\n4. Flooding\n5. Power outages",
"en_article_excerpt": "Typhoon Hinnamnor, known in the Philippines as Super Typhoon Henry, was a very large and powerful tropical cyclone that impacted Japan, South Korea, Taiwan, the Philippines, and Russia.",
"en_article_url": "https://en.wikipedia.org/wiki/Typhoon_Hinnamnor",
"event_category": "disasters",
"event_name": "typhoon_hinnamno",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Hinnamnor, known in the Philippines as Super Typhoon Henry, was a very large and powerful tropical cyclone that impacted Japan, South Korea, Taiwan, the Philippines, and Russia.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nGiven the query \"Typhoon Hinnamnor, known as Super Typhoon Henry, was a very large and powerful tropical cyclone that impacted Japan, South Korea, Taiwan, the Philippines, and Russia\", it is reasonable to extract events that involve the typhoon impacting these countries. The query specifically mentions several countries and regions where the typhoon had significant impacts, suggesting that the events should focus on these locations.\n\nEVENTS:\n1. Typhoon Hinnamnor making landfall in Japan\n2. Typhoon Hinnamnor affecting Taiwan\n3. Typhoon Hinnamnor affecting South Korea\n4. Typhoon Hinnamnor making landfall in the Philippines\n5. Typhoon Hinnamnor causing significant damage in Russia",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Hinnamnor, known in the Philippines as Super Typhoon Henry, was a very large and powerful tropical cyclone that impacted Japan, South Korea, Taiwan, the Philippines, and Russia.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nConsidering the query \"Typhoon Hinnamnor, known in the Philippines as Super Typhoon Henry\", I would extract events that are likely to be seen in a video related to this event. These events should be concrete enough to be visualized in a video, as the query is describing a natural disaster that can be observed in various locations. Based on this, the following events seem reasonable to extract:\n\n1. \"Typhoon Hinnamnor\" \n2. \"Super Typhoon Henry\" \n3. \"Impact on Japan\" \n4. \"Impact on Korea\" \n5. \"Typhoon Hinnamnor makes landfall\" \n\nEVENTS: \n1. Typhoon Hinnamnor\n2. Super Typhoon Henry\n3. Impact on Japan\n4. Impact on Korea\n5. Typhoon Hinnamnor makes landfall",
"video_URL": "http://twitter.com/i/status/1567139282846089217",
"video_description": "제11호 태풍 힌남노로 인한 자연재해 피해가 속출했는데요. 포항 아파트 지하주차장 침수로 인한 대규모 실종자가 발생을 했습니다. 포항 우방신세계아파트라는 곳에서 침수로 인해 실종자가 발생했고 9월 6일 20시 40분경 실종자1명(39세 남성)을 구조했고, 21시 50분 실종자1명(여성) 구조.",
"video_id": "1567139282846089217"
} | 16 | [
"The image shows a nighttime scene at what appears to be an outdoor pool or water attraction. A group of people is visible in the water, with some standing and others swimming. Two people in orange life vests are seen on the side of the pool, possibly assisting or supervising. A crowd is gathered on a balcony above... | 1567139282846089217 | A nighttime event at an outdoor water attraction appears to be a rescue operation. The scene features a group of people in orange life vests assisting someone in the water, with rescue lights illuminating the area. The rescuers are providing medical assistance, and the scene is set in a parking area near an outdoor water attraction. |
Typhoon Hinnamnor, known in the Philippines as Super Typhoon Henry, was a very large and powerful tropical cyclone that impacted Japan, South Korea, Taiwan, the Philippines, and Russia. | [
"\"Typhoon Hinnamnor making landfall in Japan around [insert time]\"",
"\"Typhoon Hinnamnor in Japan at time NOT AVAILABLE\"",
"\"Typhoon Hinnamnor in South Korea\"",
"\"Typhoon Hinnamnor making landfall in Taiwan in [time] in Japan\"",
"typhoon hinnamnor weather forecast philippines",
"\"Hinamnor Typhoon... | [
"\"Superteon\" \"Hinnamnor\"",
"typhoon hinnamnor japan at 14:30 this weekend",
"weather forecast north korea typhoon",
"\"Typhoon Hinnamnor in Taiwan 2024\"",
"typhoon hinnamnor typhoon hinnamnor philippines 1989",
"\"Typhoon Hinnamnor in Russia at unknown time\"",
"\"Super Typhoon Hinnamnor Superteon ... | [
"Typhoon Hinnamnor/Superteon Henry",
"Japan Landfall",
"\"South Korea landfall during [time]\"",
"\"Taiwan\" landfall",
"landfall philippines",
"\"Russia landfall [time]\"",
"Typhoon Hinnamnor/Superteon Henry Widespread destruction",
"Japan: Wildfires",
"\"South Korea flood 2017\"",
"\"Taiwan natu... | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%83%9C%ED%92%8D_%ED%9E%8C%EB%82%A8%EB%85%B8",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Hinnamnor, known in the Philippines as Super Typhoon Henry, was a very large and powerful tropical cyclone that impacted Japan, South Korea, Taiwan, the Philippines, and Russia.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nIt is likely to extract the following events from the given query:\n \nThese events seem to be related to the topic of Typhoon Hinnamnor in its impact on other countries:\n1. Typhoon Hinnamnor makes landfall in Japan, South Korea, Taiwan, the Philippines, and Russia.\n2. The storm causes widespread destruction, landslides and flooding in the Philippines.\n3. The storm brings heavy rainfall and strong winds that cause massive power outages in the region.\n4. The international community issues warnings about the potential for widespread damage and loss of life.\n\nEVENTS:\n1. Landfall\n2. Widespread destruction\n3. Landslides\n4. Flooding\n5. Power outages",
"en_article_excerpt": "Typhoon Hinnamnor, known in the Philippines as Super Typhoon Henry, was a very large and powerful tropical cyclone that impacted Japan, South Korea, Taiwan, the Philippines, and Russia.",
"en_article_url": "https://en.wikipedia.org/wiki/Typhoon_Hinnamnor",
"event_category": "disasters",
"event_name": "typhoon_hinnamno",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Hinnamnor, known in the Philippines as Super Typhoon Henry, was a very large and powerful tropical cyclone that impacted Japan, South Korea, Taiwan, the Philippines, and Russia.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nGiven the query \"Typhoon Hinnamnor, known as Super Typhoon Henry, was a very large and powerful tropical cyclone that impacted Japan, South Korea, Taiwan, the Philippines, and Russia\", it is reasonable to extract events that involve the typhoon impacting these countries. The query specifically mentions several countries and regions where the typhoon had significant impacts, suggesting that the events should focus on these locations.\n\nEVENTS:\n1. Typhoon Hinnamnor making landfall in Japan\n2. Typhoon Hinnamnor affecting Taiwan\n3. Typhoon Hinnamnor affecting South Korea\n4. Typhoon Hinnamnor making landfall in the Philippines\n5. Typhoon Hinnamnor causing significant damage in Russia",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Hinnamnor, known in the Philippines as Super Typhoon Henry, was a very large and powerful tropical cyclone that impacted Japan, South Korea, Taiwan, the Philippines, and Russia.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nConsidering the query \"Typhoon Hinnamnor, known in the Philippines as Super Typhoon Henry\", I would extract events that are likely to be seen in a video related to this event. These events should be concrete enough to be visualized in a video, as the query is describing a natural disaster that can be observed in various locations. Based on this, the following events seem reasonable to extract:\n\n1. \"Typhoon Hinnamnor\" \n2. \"Super Typhoon Henry\" \n3. \"Impact on Japan\" \n4. \"Impact on Korea\" \n5. \"Typhoon Hinnamnor makes landfall\" \n\nEVENTS: \n1. Typhoon Hinnamnor\n2. Super Typhoon Henry\n3. Impact on Japan\n4. Impact on Korea\n5. Typhoon Hinnamnor makes landfall",
"video_URL": "http://twitter.com/i/status/1566015450982494209",
"video_description": "얘들아… 초속 54미터 이상이면 건물도 붕괴될 정도의 바람이라는데 이번 힌남노는 매미보다 중심 기압이 낮은 더 강력한 태풍이라 제주/남부해안지역엔 초속 60미터까지도 예상된대.. 영동 쪽은 초속 3~40미터 수도권은 초속 15미터 예상… 다들 진짜 조심하자 대비할 수 있는 부분은 꼭 하고ㅜ",
"video_id": "1566015450982494209"
} | 16 | [
"In the image, two news anchors are seated at a desk in a studio setting, discussing a topic. The background shows a large screen displaying a powerful wave crashing against rocks, indicating a severe weather condition. The text on the screen reads \"기존 역대 1위 바람 태풍 '메미'…제주도에서 초속 60m 기록,\" which translates to \"Prev... | 1566015450982494209 | A severe weather event, likely a typhoon, has occurred in Korea, causing widespread disruption and danger to various areas, including coastal regions. The storm, 'Hinnamnor', brought strong winds and rainfall, prompting a weather advisory warning of potential storm surges and flooding. Residents are advised to take necessary precautions to ensure safety. The storm's impact has been felt in multiple locations, with residents in Jeju Island experiencing record-breaking wind speeds and typhoon surges, as well as flooding in other parts of the country. |
Typhoon Hinnamnor, known in the Philippines as Super Typhoon Henry, was a very large and powerful tropical cyclone that impacted Japan, South Korea, Taiwan, the Philippines, and Russia. | [
"\"Typhoon Hinnamnor making landfall in Japan around [insert time]\"",
"\"Typhoon Hinnamnor in Japan at time NOT AVAILABLE\"",
"\"Typhoon Hinnamnor in South Korea\"",
"\"Typhoon Hinnamnor making landfall in Taiwan in [time] in Japan\"",
"typhoon hinnamnor weather forecast philippines",
"\"Hinamnor Typhoon... | [
"\"Superteon\" \"Hinnamnor\"",
"typhoon hinnamnor japan at 14:30 this weekend",
"weather forecast north korea typhoon",
"\"Typhoon Hinnamnor in Taiwan 2024\"",
"typhoon hinnamnor typhoon hinnamnor philippines 1989",
"\"Typhoon Hinnamnor in Russia at unknown time\"",
"\"Super Typhoon Hinnamnor Superteon ... | [
"Typhoon Hinnamnor/Superteon Henry",
"Japan Landfall",
"\"South Korea landfall during [time]\"",
"\"Taiwan\" landfall",
"landfall philippines",
"\"Russia landfall [time]\"",
"Typhoon Hinnamnor/Superteon Henry Widespread destruction",
"Japan: Wildfires",
"\"South Korea flood 2017\"",
"\"Taiwan natu... | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%83%9C%ED%92%8D_%ED%9E%8C%EB%82%A8%EB%85%B8",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Hinnamnor, known in the Philippines as Super Typhoon Henry, was a very large and powerful tropical cyclone that impacted Japan, South Korea, Taiwan, the Philippines, and Russia.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nIt is likely to extract the following events from the given query:\n \nThese events seem to be related to the topic of Typhoon Hinnamnor in its impact on other countries:\n1. Typhoon Hinnamnor makes landfall in Japan, South Korea, Taiwan, the Philippines, and Russia.\n2. The storm causes widespread destruction, landslides and flooding in the Philippines.\n3. The storm brings heavy rainfall and strong winds that cause massive power outages in the region.\n4. The international community issues warnings about the potential for widespread damage and loss of life.\n\nEVENTS:\n1. Landfall\n2. Widespread destruction\n3. Landslides\n4. Flooding\n5. Power outages",
"en_article_excerpt": "Typhoon Hinnamnor, known in the Philippines as Super Typhoon Henry, was a very large and powerful tropical cyclone that impacted Japan, South Korea, Taiwan, the Philippines, and Russia.",
"en_article_url": "https://en.wikipedia.org/wiki/Typhoon_Hinnamnor",
"event_category": "disasters",
"event_name": "typhoon_hinnamno",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Hinnamnor, known in the Philippines as Super Typhoon Henry, was a very large and powerful tropical cyclone that impacted Japan, South Korea, Taiwan, the Philippines, and Russia.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nGiven the query \"Typhoon Hinnamnor, known as Super Typhoon Henry, was a very large and powerful tropical cyclone that impacted Japan, South Korea, Taiwan, the Philippines, and Russia\", it is reasonable to extract events that involve the typhoon impacting these countries. The query specifically mentions several countries and regions where the typhoon had significant impacts, suggesting that the events should focus on these locations.\n\nEVENTS:\n1. Typhoon Hinnamnor making landfall in Japan\n2. Typhoon Hinnamnor affecting Taiwan\n3. Typhoon Hinnamnor affecting South Korea\n4. Typhoon Hinnamnor making landfall in the Philippines\n5. Typhoon Hinnamnor causing significant damage in Russia",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Hinnamnor, known in the Philippines as Super Typhoon Henry, was a very large and powerful tropical cyclone that impacted Japan, South Korea, Taiwan, the Philippines, and Russia.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nConsidering the query \"Typhoon Hinnamnor, known in the Philippines as Super Typhoon Henry\", I would extract events that are likely to be seen in a video related to this event. These events should be concrete enough to be visualized in a video, as the query is describing a natural disaster that can be observed in various locations. Based on this, the following events seem reasonable to extract:\n\n1. \"Typhoon Hinnamnor\" \n2. \"Super Typhoon Henry\" \n3. \"Impact on Japan\" \n4. \"Impact on Korea\" \n5. \"Typhoon Hinnamnor makes landfall\" \n\nEVENTS: \n1. Typhoon Hinnamnor\n2. Super Typhoon Henry\n3. Impact on Japan\n4. Impact on Korea\n5. Typhoon Hinnamnor makes landfall",
"video_URL": "http://twitter.com/i/status/1566948535899041792",
"video_description": "칭구가 보내준 집앞 영상 괜찮아??? 응 근데 나도 안괜찮아 다들 괜찮으시길 #태풍_힌남노_저리가",
"video_id": "1566948535899041792"
} | 16 | [
"The image shows a large landslide on a hillside with significant soil and debris displacement. In the foreground, there's a modern building with a flat roof and a car parked nearby. Behind the landslide, a row of multi-story buildings can be seen, indicating a residential area. The surrounding area is green with t... | 1566948535899041792 | A significant landslide occurred in an urban area, causing destruction of buildings and displacement of soil and debris. The event was part of ongoing rain, which may have played a role in exacerbating the landslide. |
Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019. | [
"typhoon lingling",
"Typhoon Lingling",
"\"Typhoon Lingling in Korea\""
] | [
"I don't have any information about the event, place, or time, but I can search for infrastructure destruction in relation to typhoons",
"\"Typhoon-related weather alerts and flood warnings in storm-prone regions\"",
"\"Typhoon Lingling coastal erosion, high waves, in the Philippines\"",
"\"Typhoon Lingling c... | [
"Typhoon Lingling",
"\"Typhoon Lingling in Taiwan, eastern region, causing widespread destruction and loss of life\"",
"typhoon Philippines",
"\"Typhoon Lingling in Taiwan\"",
"\"Typhoon Lingling in Korea in [time] affected the region with significant flooding due to strong winds.\""
] | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%83%9C%ED%92%8D_%EB%A7%81%EB%A7%81_(2019%EB%85%84)",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nTo extract meaningful events from the given query, we need to identify the specific and concrete information that could be visualized in a video. Since the query is about a typhoon, we can extract events related to the storm and its impact on the region.\n\nEVENTS:\n1. Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea.\n2. Typhoon Lingling made landfall in Taiwan, causing widespread destruction.\n3. The typhoon brought heavy rain and strong winds to the Philippines.\n4. The storm caused widespread power outages in Taiwan.\n5. The typhoon's strong winds and heavy rain led to flooding in Korea.",
"en_article_excerpt": "Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019.",
"en_article_url": "https://en.wikipedia.org/wiki/Typhoon_Lingling_(2019)",
"event_category": "disasters",
"event_name": "typhoon_lingling",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nGiven the query \"Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019\", it makes sense to extract events that are specific and concrete enough to be visualized in a video, such as typhoons, Philippines, China, and Korea.\n\nEVENTS:\n1. Typhoon Lingling hitting the Philippines\n2. Typhoon Liwayway hitting China\n3. Typhoon Lingling hitting Korea",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nGiven the query \"Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019\", the events that are likely to be visualized in a video are related to the typhoon. The query implies that the typhoon is an ongoing event, so we should consider events that could be related to its impact, destruction, or aftermath.\n\nEVENTS:\n1. Destruction of infrastructure, such as buildings, roads, and bridges\n2. Flooding, including storm surges, high tides, and heavy rainfall\n3. Storm surges, high waves, and coastal erosion\n4. Loss of life, injuries, and displacement of people\n5. Destruction of crops, livestock, and property",
"video_URL": "http://youtube.com/watch?v=QJY4XlSNVf8",
"video_description": "#태풍 #링링 #제13호 #13호 #중국 (서울=연합뉴스) 제13호 태풍 '링링'이 세력을 키우며 한반도를 향해 북상하고 있습니다. 5일 기상청에 따르면 '링링'은 이날 오전 9시 현재 일본 오키나와 남서쪽 약 370㎞ 해상에서 시속 19㎞로 북상 중입니다. 발생 초기 강도가 '약'이던 '링링'은 수온이 높은 해역을 지나며 이날 현재 '강'으로 세력이 강해진 상태입니다. 한국과 미국, 일본 기상청 모두 '링링'이 토요일인 7일 수도권을 강타할 것으로 예보해 철저한 대비가 필요합니다. 중국 저장성과 하이난성에서는 지난 2일 이후 '링링'과 제14호 태풍 '가지키'의 영향으로 폭우 피해가 잇따랐는데요. 영상으로 보시죠. ◆ 연합뉴스 홈페이지→ http://www.yna.co.kr/ ◆ 이 시각 많이 본 기사 → https://goo.gl/VQTsSZ ◆ 오늘의 핫뉴스 → https://goo.gl/WyGXpG ◆ 현장영상 → https://goo.gl/5aZcx8 ◆ 카드뉴스 →https://goo.gl/QKfDTH ◆연합뉴스 공식 SNS◆ ◇페이스북→ https://www.facebook.com/yonhap/ ▣ 연합뉴스 유튜브 채널 구독 : https://goo.gl/pL7TmT ▣ 연합뉴스 인스타 : https://goo.gl/UbqiQb",
"video_id": "QJY4XlSNVf8"
} | 16 | [
"The image depicts a rocky shoreline with large waves crashing against the rocks. The scene appears stormy, with dark clouds in the background. The text in the image, which is in Korean, translates to \"One-hit knockout wave in 13 seconds 'Tipping'.\"",
"The video shows a heavy rainstorm in an urban street settin... | QJY4XlSNVf8 | A severe typhoon (Typhoon Lingling and Typhoon Kajiki) caused widespread flooding in various areas of South Korea, particularly in Haman City, as well as in the Seoul metropolitan area. The typhoon affected homes, transportation, and businesses, leading to damage and disruption to daily life. The flooding was exacerbated by strong winds, heavy rainfall, and the storm's rapid movement, causing widespread flooding and displacement of residents.
The typhoon's impact was reported by local news agencies, including Yonhap News Agency, and was anticipated to continue affecting the region for several days. The flooding was attributed to the typhoon's influence on the southwestern sea of Jeju Island, where it passed, causing the storm to affect the Seoul metropolitan area and surrounding regions.
The flooding was reported in various locations, including streets, homes, and businesses, with the most affected areas being Haman City and the Seoul metropolitan area. The disaster was exacerbated by the typhoon's strong winds, heavy rainfall, and its rapid movement, leading to a rapid rise in water levels and displacement of residents.
The disaster was reported to have caused significant damage to infrastructure, homes, and businesses, and left many residents without access to basic necessities like food, water, and electricity. The situation was considered severe, with the rapid movement of the typhoon causing widespread flooding and displacement.
The typhoon was predicted to continue to affect the region for several days, with the Seoul metropolitan area being particularly vulnerable to its impact. The disaster was attributed to the typhoon's influence on the southwestern sea of Jeju Island, where it passed, causing the storm to affect the region. |
Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019. | [
"typhoon lingling",
"Typhoon Lingling",
"\"Typhoon Lingling in Korea\""
] | [
"I don't have any information about the event, place, or time, but I can search for infrastructure destruction in relation to typhoons",
"\"Typhoon-related weather alerts and flood warnings in storm-prone regions\"",
"\"Typhoon Lingling coastal erosion, high waves, in the Philippines\"",
"\"Typhoon Lingling c... | [
"Typhoon Lingling",
"\"Typhoon Lingling in Taiwan, eastern region, causing widespread destruction and loss of life\"",
"typhoon Philippines",
"\"Typhoon Lingling in Taiwan\"",
"\"Typhoon Lingling in Korea in [time] affected the region with significant flooding due to strong winds.\""
] | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%83%9C%ED%92%8D_%EB%A7%81%EB%A7%81_(2019%EB%85%84)",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nTo extract meaningful events from the given query, we need to identify the specific and concrete information that could be visualized in a video. Since the query is about a typhoon, we can extract events related to the storm and its impact on the region.\n\nEVENTS:\n1. Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea.\n2. Typhoon Lingling made landfall in Taiwan, causing widespread destruction.\n3. The typhoon brought heavy rain and strong winds to the Philippines.\n4. The storm caused widespread power outages in Taiwan.\n5. The typhoon's strong winds and heavy rain led to flooding in Korea.",
"en_article_excerpt": "Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019.",
"en_article_url": "https://en.wikipedia.org/wiki/Typhoon_Lingling_(2019)",
"event_category": "disasters",
"event_name": "typhoon_lingling",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nGiven the query \"Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019\", it makes sense to extract events that are specific and concrete enough to be visualized in a video, such as typhoons, Philippines, China, and Korea.\n\nEVENTS:\n1. Typhoon Lingling hitting the Philippines\n2. Typhoon Liwayway hitting China\n3. Typhoon Lingling hitting Korea",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nGiven the query \"Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019\", the events that are likely to be visualized in a video are related to the typhoon. The query implies that the typhoon is an ongoing event, so we should consider events that could be related to its impact, destruction, or aftermath.\n\nEVENTS:\n1. Destruction of infrastructure, such as buildings, roads, and bridges\n2. Flooding, including storm surges, high tides, and heavy rainfall\n3. Storm surges, high waves, and coastal erosion\n4. Loss of life, injuries, and displacement of people\n5. Destruction of crops, livestock, and property",
"video_URL": "http://youtube.com/watch?v=ZFR-SBQwtTw",
"video_description": "태풍 '링링' 발생…주후반 한반도 영향 가능성 오늘 오전 9시쯤 필리핀 동쪽 해상에서 13호 태풍 '링링'이 발생했습니다. 강풍반경 200km의 소형 태풍인 '링링'은 계속 북상해서 수요일쯤 타이완 동쪽 해상에 접근하겠습니다. 태풍 '링링'은 금요일 오전쯤 제주도 서귀포 남서쪽 390km 해상까지 진출하겠고 이후 서해상을 향할 것으로 보입니다. 기상청은 태풍 발생 초기라 경로가 유동적이지만 주 후반부터 한반도 부근으로 북상할 가능성이 크다고 내다봤습니다. 연합뉴스TV 기사문의 및 제보 : 카톡/라인 jebo23 ▣ 연합뉴스TV 유튜브 채널 구독 https://goo.gl/VuCJMi ▣ 대한민국 뉴스의 시작 연합뉴스TV / Yonhap News TV http://www.yonhapnewstv.co.kr/",
"video_id": "ZFR-SBQwtTw"
} | 16 | [
"In this frame, a news broadcast is taking place with two anchors seated at a desk. The setting is a news studio with multiple monitors and control equipment in the background. The text on the screen includes news headlines in Korean and a ticker displaying additional information. The broadcast appears to be from Y... | ZFR-SBQwtTw | A severe storm, identified as Typhoon 'Ringling', is affecting the Korean Peninsula, with strong winds causing damage to palm trees, coastal areas, and infrastructure. The storm surge is leading to flooding and potential damage to homes and buildings. The storm's impact is exacerbated by high tides and strong winds, with waves crashing against coastal areas and generating large waves. A group of people, possibly engineers or responders, are seen attempting to protect themselves and their belongings, while a lighthouse stands as a warning of the storm's severity. The storm continues to pose a threat to the region, with Typhoon 'Ringling' expected to make landfall in the second half of the week, leading to further damage and displacement. |
Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019. | [
"typhoon lingling",
"Typhoon Lingling",
"\"Typhoon Lingling in Korea\""
] | [
"I don't have any information about the event, place, or time, but I can search for infrastructure destruction in relation to typhoons",
"\"Typhoon-related weather alerts and flood warnings in storm-prone regions\"",
"\"Typhoon Lingling coastal erosion, high waves, in the Philippines\"",
"\"Typhoon Lingling c... | [
"Typhoon Lingling",
"\"Typhoon Lingling in Taiwan, eastern region, causing widespread destruction and loss of life\"",
"typhoon Philippines",
"\"Typhoon Lingling in Taiwan\"",
"\"Typhoon Lingling in Korea in [time] affected the region with significant flooding due to strong winds.\""
] | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%83%9C%ED%92%8D_%EB%A7%81%EB%A7%81_(2019%EB%85%84)",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nTo extract meaningful events from the given query, we need to identify the specific and concrete information that could be visualized in a video. Since the query is about a typhoon, we can extract events related to the storm and its impact on the region.\n\nEVENTS:\n1. Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea.\n2. Typhoon Lingling made landfall in Taiwan, causing widespread destruction.\n3. The typhoon brought heavy rain and strong winds to the Philippines.\n4. The storm caused widespread power outages in Taiwan.\n5. The typhoon's strong winds and heavy rain led to flooding in Korea.",
"en_article_excerpt": "Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019.",
"en_article_url": "https://en.wikipedia.org/wiki/Typhoon_Lingling_(2019)",
"event_category": "disasters",
"event_name": "typhoon_lingling",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nGiven the query \"Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019\", it makes sense to extract events that are specific and concrete enough to be visualized in a video, such as typhoons, Philippines, China, and Korea.\n\nEVENTS:\n1. Typhoon Lingling hitting the Philippines\n2. Typhoon Liwayway hitting China\n3. Typhoon Lingling hitting Korea",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nGiven the query \"Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019\", the events that are likely to be visualized in a video are related to the typhoon. The query implies that the typhoon is an ongoing event, so we should consider events that could be related to its impact, destruction, or aftermath.\n\nEVENTS:\n1. Destruction of infrastructure, such as buildings, roads, and bridges\n2. Flooding, including storm surges, high tides, and heavy rainfall\n3. Storm surges, high waves, and coastal erosion\n4. Loss of life, injuries, and displacement of people\n5. Destruction of crops, livestock, and property",
"video_URL": "http://youtube.com/watch?v=8-DXe4qYo2E",
"video_description": "#태풍 #링링 (서울=연합뉴스) 제13호 태풍 '링링'이 필리핀 동부에서 발생했습니다. 이 태풍의 중심기압은 1천hPa(헥토파스칼), 중심 부근 최대 풍속은 초속 18ｍ(시속 65㎞)이며, 강풍 반경은 200㎞입니다. 한반도에 접근할 무렵 지금보다 크고 강해질 것으로 전망돼 대비가 필요한데요. 자세한 내용은 영상으로 보시죠. ◆ 연합뉴스 홈페이지→ http://www.yna.co.kr/ ◆ 이 시각 많이 본 기사 → https://goo.gl/VQTsSZ ◆ 오늘의 핫뉴스 → https://goo.gl/WyGXpG ◆ 현장영상 → https://goo.gl/5aZcx8 ◆ 카드뉴스 →https://goo.gl/QKfDTH ◆연합뉴스 공식 SNS◆ ◇페이스북→ https://www.facebook.com/yonhap/ ▣ 연합뉴스 유튜브 채널 구독 : https://goo.gl/pL7TmT ▣ 연합뉴스 인스타 : https://goo.gl/UbqiQb",
"video_id": "8-DXe4qYo2E"
} | 16 | [
"The image shows a weather map depicting the path of a tropical storm or typhoon. The storm is labeled as \"呤呤\" and is currently at the 13th category. The map highlights the storm's projected trajectory over the Philippines. The image includes text in Korean, indicating that the storm has formed in the eastern Pac... | 8-DXe4qYo2E | The video is of a typhoon that has formed in the eastern Pacific and is projected to affect the Philippines. The storm is intensifying, moving towards the Korean peninsula, and will reach its peak intensity in 6-7 days. The storm's projected path, with waves and strong winds of up to 18 meters per second, has caused significant destruction and displacement in the affected areas. The storm has already caused widespread flooding, and its movement will continue to worsen the situation. The storm is expected to pass south of Seogwipo, Jeju Island, this week, with its impact felt in the surrounding areas. The weather system is expected to improve as it moves away from the peninsula, with heavy rain and strong winds dissipating as it exits. The storm has already caused a significant disruption, with many residents affected, including vehicles, roads, and buildings. |
Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019. | [
"typhoon lingling",
"Typhoon Lingling",
"\"Typhoon Lingling in Korea\""
] | [
"I don't have any information about the event, place, or time, but I can search for infrastructure destruction in relation to typhoons",
"\"Typhoon-related weather alerts and flood warnings in storm-prone regions\"",
"\"Typhoon Lingling coastal erosion, high waves, in the Philippines\"",
"\"Typhoon Lingling c... | [
"Typhoon Lingling",
"\"Typhoon Lingling in Taiwan, eastern region, causing widespread destruction and loss of life\"",
"typhoon Philippines",
"\"Typhoon Lingling in Taiwan\"",
"\"Typhoon Lingling in Korea in [time] affected the region with significant flooding due to strong winds.\""
] | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%83%9C%ED%92%8D_%EB%A7%81%EB%A7%81_(2019%EB%85%84)",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nTo extract meaningful events from the given query, we need to identify the specific and concrete information that could be visualized in a video. Since the query is about a typhoon, we can extract events related to the storm and its impact on the region.\n\nEVENTS:\n1. Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea.\n2. Typhoon Lingling made landfall in Taiwan, causing widespread destruction.\n3. The typhoon brought heavy rain and strong winds to the Philippines.\n4. The storm caused widespread power outages in Taiwan.\n5. The typhoon's strong winds and heavy rain led to flooding in Korea.",
"en_article_excerpt": "Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019.",
"en_article_url": "https://en.wikipedia.org/wiki/Typhoon_Lingling_(2019)",
"event_category": "disasters",
"event_name": "typhoon_lingling",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nGiven the query \"Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019\", it makes sense to extract events that are specific and concrete enough to be visualized in a video, such as typhoons, Philippines, China, and Korea.\n\nEVENTS:\n1. Typhoon Lingling hitting the Philippines\n2. Typhoon Liwayway hitting China\n3. Typhoon Lingling hitting Korea",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nGiven the query \"Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019\", the events that are likely to be visualized in a video are related to the typhoon. The query implies that the typhoon is an ongoing event, so we should consider events that could be related to its impact, destruction, or aftermath.\n\nEVENTS:\n1. Destruction of infrastructure, such as buildings, roads, and bridges\n2. Flooding, including storm surges, high tides, and heavy rainfall\n3. Storm surges, high waves, and coastal erosion\n4. Loss of life, injuries, and displacement of people\n5. Destruction of crops, livestock, and property",
"video_URL": "http://youtube.com/watch?v=taOktWIvk20",
"video_description": "#태풍 #링링 #서울태풍 #수도권태풍 #최근접 #초강력태풍 (서울=연합뉴스) 한반도를 강타할 초강력 제13호 태풍 '링링'이 왔습니다. 기상청에 따르면 태풍은 7일 새벽 서귀포 서쪽 약 150㎞ 해상을 통과, 오전 6시께 전남 목포 서쪽 약 140㎞ 해상을 지났습니다. 정오께 서울 서남서쪽 약 140㎞ 해상을 지날 것으로 보이는데요. 오후 2시께 태풍 중심이 서울에 제일 가까울 전망입니다. 영상으로 보시겠습니다. [영상 : 연합뉴스TV] ◆ 연합뉴스 홈페이지→ http://www.yna.co.kr/ ◆ 이 시각 많이 본 기사 → https://goo.gl/VQTsSZ ◆ 오늘의 핫뉴스 → https://goo.gl/WyGXpG ◆ 현장영상 → https://goo.gl/5aZcx8 ◆ 카드뉴스 →https://goo.gl/QKfDTH ◆연합뉴스 공식 SNS◆ ◇페이스북→ https://www.facebook.com/yonhap/ ▣ 연합뉴스 유튜브 채널 구독 : https://goo.gl/pL7TmT ▣ 연합뉴스 인스타 : https://goo.gl/UbqiQb",
"video_id": "taOktWIvk20"
} | 16 | [
"The image shows a street scene with trees and power lines. The visibility is reduced due to what appears to be a mist or light rain. There is a text overlay in Korean at the bottom of the image, which translates to \"A typhoon causing damage to Han River, Typhoon Rinkheng 13th\".",
"The image continues to show a... | taOktWIvk20 | A typhoon, Typhoon Rinkheng 13th, caused significant damage to buildings and infrastructure in the Han River area, particularly in the port city of Seogwipo, Jeju. The storm resulted in destroyed roofs, damaged roads, and hazardous conditions for maritime travel. The storm's impact was further exacerbated by strong winds and rain. Authorities implemented special warnings and took measures to secure infrastructure and ensure public safety. A typhoon warning was issued for the area around 140km from Seoul, and the storm's trajectory was expected to dissipate by noon. |
Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019. | [
"typhoon lingling",
"Typhoon Lingling",
"\"Typhoon Lingling in Korea\""
] | [
"I don't have any information about the event, place, or time, but I can search for infrastructure destruction in relation to typhoons",
"\"Typhoon-related weather alerts and flood warnings in storm-prone regions\"",
"\"Typhoon Lingling coastal erosion, high waves, in the Philippines\"",
"\"Typhoon Lingling c... | [
"Typhoon Lingling",
"\"Typhoon Lingling in Taiwan, eastern region, causing widespread destruction and loss of life\"",
"typhoon Philippines",
"\"Typhoon Lingling in Taiwan\"",
"\"Typhoon Lingling in Korea in [time] affected the region with significant flooding due to strong winds.\""
] | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%83%9C%ED%92%8D_%EB%A7%81%EB%A7%81_(2019%EB%85%84)",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nTo extract meaningful events from the given query, we need to identify the specific and concrete information that could be visualized in a video. Since the query is about a typhoon, we can extract events related to the storm and its impact on the region.\n\nEVENTS:\n1. Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea.\n2. Typhoon Lingling made landfall in Taiwan, causing widespread destruction.\n3. The typhoon brought heavy rain and strong winds to the Philippines.\n4. The storm caused widespread power outages in Taiwan.\n5. The typhoon's strong winds and heavy rain led to flooding in Korea.",
"en_article_excerpt": "Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019.",
"en_article_url": "https://en.wikipedia.org/wiki/Typhoon_Lingling_(2019)",
"event_category": "disasters",
"event_name": "typhoon_lingling",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nGiven the query \"Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019\", it makes sense to extract events that are specific and concrete enough to be visualized in a video, such as typhoons, Philippines, China, and Korea.\n\nEVENTS:\n1. Typhoon Lingling hitting the Philippines\n2. Typhoon Liwayway hitting China\n3. Typhoon Lingling hitting Korea",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nGiven the query \"Typhoon Lingling, also known in the Philippines as Typhoon Liwayway, was a powerful typhoon that hit the Philippines, China, and Korea in 2019\", the events that are likely to be visualized in a video are related to the typhoon. The query implies that the typhoon is an ongoing event, so we should consider events that could be related to its impact, destruction, or aftermath.\n\nEVENTS:\n1. Destruction of infrastructure, such as buildings, roads, and bridges\n2. Flooding, including storm surges, high tides, and heavy rainfall\n3. Storm surges, high waves, and coastal erosion\n4. Loss of life, injuries, and displacement of people\n5. Destruction of crops, livestock, and property",
"video_URL": "http://youtube.com/watch?v=t1uezBvD9U8",
"video_description": "태풍 '링링'이 서해안을 따라 빠르게 북상하면서 충북 지역 피해도 속출하고 있습니다. 충북 지역 피해 현장을 카메라에 담았습니다.",
"video_id": "t1uezBvD9U8"
} | 16 | [
"The current frame depicts a large tree that has fallen, with its roots and branches exposed on the sidewalk. A metal circular structure, possibly part of a playground or street furniture, is entangled in the tree's roots. The scene appears to be in an urban area with buildings and a sign that reads \"PARIS\" in th... | t1uezBvD9U8 | A fallen tree, originating from a playground or street furniture, causes significant disruption in an urban area with a "PARIS" sign and "No Parking" sign. The tree's roots and branches obstruct traffic, while a person in a yellow vest assesses the situation. The scene shows a large, tilted billboard that is secured with police tape, obstructing the sidewalk. Ongoing cleanup efforts are visible, with debris scattered on the ground and broken objects present. The area is cordoned off to maintain public safety. |
Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012. | [
"Typhoon Soulik severe weather New York",
"Typhoon Soulik, USA, not available",
"typhoon Soulik New York",
"typhoon soulik usa not available time",
"\"Typhoon Soulik New York\"",
"\"Typhoon Soulik in the USA during NOT available time\"",
"Typhoon Soulik strikes New York, potentially causing widespread d... | [
"\"Typhoon Soulik in New York around [time period]\"",
"typhoon soulik southeast usa at this time",
"\"New York typhoon Soulik strikes\"",
"\"Typhoon Soulik in the USA for [insert time frame, e.g., '2020-2021']\"",
"\"Typhoon affecting New York in September\"",
"\"typhoon landfall in usa\"",
"\"Typhoon ... | [
"New York in Typhoon Soulik",
"\"typhoon soulik hits coastal areas in usa at what time\"",
"\"Typhoon Soulik in New York causing widespread flooding\"",
"typhoon -usa",
"\"New York in the path of typhoon Soulik\"",
"Typhoon Soulik storm surge, USA",
"\"New York during typhoon Soulik\"",
"\"Typhoon Sou... | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%83%9C%ED%92%8D_%EC%86%94%EB%A6%AD_(2018%EB%85%84)",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nSince \"Typhoon Soulik\" is a specific typhoon, it makes sense to extract events related to the storm itself or its impact. The query does not mention any human-related events or any other context, which suggests that extracting the typhoon itself and its effects would be a good idea.\n\nEVENTS:\n1. The storm hits a coastal area\n2. The storm causes widespread flooding\n3. A massive storm surge is generated by the typhoon\n4. Thousands of people are affected by the storm\n5. The storm causes widespread damage to infrastructure",
"en_article_excerpt": "Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.",
"en_article_url": "https://en.wikipedia.org/wiki/Typhoon_Soulik_(2018)",
"event_category": "disasters",
"event_name": "typhoon_soulik",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThe given query is a search query for Typhoon Soulik. It's likely that it could be visualized in a video as a destructive event, possibly a devastating storm with severe damage, given the description. Considering this, we should focus on events related to severe weather, destruction, and disaster.\n\nEVENTS:\n1. Severe weather\n2. Destruction\n3. Disaster\n4. Natural disaster",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nThe query seems to be asking for events that are likely to be associated with Typhoon Soulik, which made landfall in Korea. Given the mention of \"the deadliest typhoon to strike the Korean Peninsula\" and \"since Khanun in 2012\", it is reasonable to assume that Typhoon Soulik is likely to be the event in question. Therefore, the event that makes sense to extract is the typhoon itself.\n\nEVENTS:\n1. Typhoon Soulik\n2. Strike\n3. Landfall\n4. Strike\n5. Landfall",
"video_URL": "http://twitter.com/i/status/1032900418777407489",
"video_description": "제주도 태풍 솔릭의 강풍의 무서움.. 오토바이도 날려버리는 태풍 솔릭.. ㄷㄷㄷ",
"video_id": "1032900418777407489"
} | 16 | [
"A person is driving a car during a heavy snowstorm, with limited visibility outside the windshield.",
"The driver continues navigating the car in the ongoing heavy snowstorm, with wipers in motion to clear the snow from the windshield, maintaining a cautious approach on the partially covered road.",
"The drive... | 1032900418777407489 | A person navigates a car through a heavy snowstorm while maintaining caution on a partially covered road. The driver continues to use windshield wipers to clear snow from the windshield, despite poor visibility. |
Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012. | [
"Typhoon Soulik severe weather New York",
"Typhoon Soulik, USA, not available",
"typhoon Soulik New York",
"typhoon soulik usa not available time",
"\"Typhoon Soulik New York\"",
"\"Typhoon Soulik in the USA during NOT available time\"",
"Typhoon Soulik strikes New York, potentially causing widespread d... | [
"\"Typhoon Soulik in New York around [time period]\"",
"typhoon soulik southeast usa at this time",
"\"New York typhoon Soulik strikes\"",
"\"Typhoon Soulik in the USA for [insert time frame, e.g., '2020-2021']\"",
"\"Typhoon affecting New York in September\"",
"\"typhoon landfall in usa\"",
"\"Typhoon ... | [
"New York in Typhoon Soulik",
"\"typhoon soulik hits coastal areas in usa at what time\"",
"\"Typhoon Soulik in New York causing widespread flooding\"",
"typhoon -usa",
"\"New York in the path of typhoon Soulik\"",
"Typhoon Soulik storm surge, USA",
"\"New York during typhoon Soulik\"",
"\"Typhoon Sou... | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%83%9C%ED%92%8D_%EC%86%94%EB%A6%AD_(2018%EB%85%84)",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nSince \"Typhoon Soulik\" is a specific typhoon, it makes sense to extract events related to the storm itself or its impact. The query does not mention any human-related events or any other context, which suggests that extracting the typhoon itself and its effects would be a good idea.\n\nEVENTS:\n1. The storm hits a coastal area\n2. The storm causes widespread flooding\n3. A massive storm surge is generated by the typhoon\n4. Thousands of people are affected by the storm\n5. The storm causes widespread damage to infrastructure",
"en_article_excerpt": "Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.",
"en_article_url": "https://en.wikipedia.org/wiki/Typhoon_Soulik_(2018)",
"event_category": "disasters",
"event_name": "typhoon_soulik",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThe given query is a search query for Typhoon Soulik. It's likely that it could be visualized in a video as a destructive event, possibly a devastating storm with severe damage, given the description. Considering this, we should focus on events related to severe weather, destruction, and disaster.\n\nEVENTS:\n1. Severe weather\n2. Destruction\n3. Disaster\n4. Natural disaster",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nThe query seems to be asking for events that are likely to be associated with Typhoon Soulik, which made landfall in Korea. Given the mention of \"the deadliest typhoon to strike the Korean Peninsula\" and \"since Khanun in 2012\", it is reasonable to assume that Typhoon Soulik is likely to be the event in question. Therefore, the event that makes sense to extract is the typhoon itself.\n\nEVENTS:\n1. Typhoon Soulik\n2. Strike\n3. Landfall\n4. Strike\n5. Landfall",
"video_URL": "http://twitter.com/i/status/1032211760315420675",
"video_description": "실시간 제주도 서귀포 상황 태풍 솔릭...위험행..........조심하세요다들",
"video_id": "1032211760315420675"
} | 16 | [
"The current frame shows a group of people standing around a large, fallen palm tree. The scene appears to be in an urban area with several vehicles parked nearby. The fallen tree is obstructing a road, and the people seem to be assessing the situation or preparing to clear the debris. The sky is overcast, suggesti... | 1032211760315420675 | A group of people gather around a fallen palm tree in an urban area, attempting to navigate around the obstruction caused by the tree. The group consists of several vehicles, including cars and a motorcycle. The situation is depicted as a traffic jam due to the fallen tree blocking the road. Inclement weather, possibly inclement weather, is suggested by the overcast sky and the presence of rain. The people involved in the scene are not visible, suggesting they have moved out of the frame. |
Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012. | [
"Typhoon Soulik severe weather New York",
"Typhoon Soulik, USA, not available",
"typhoon Soulik New York",
"typhoon soulik usa not available time",
"\"Typhoon Soulik New York\"",
"\"Typhoon Soulik in the USA during NOT available time\"",
"Typhoon Soulik strikes New York, potentially causing widespread d... | [
"\"Typhoon Soulik in New York around [time period]\"",
"typhoon soulik southeast usa at this time",
"\"New York typhoon Soulik strikes\"",
"\"Typhoon Soulik in the USA for [insert time frame, e.g., '2020-2021']\"",
"\"Typhoon affecting New York in September\"",
"\"typhoon landfall in usa\"",
"\"Typhoon ... | [
"New York in Typhoon Soulik",
"\"typhoon soulik hits coastal areas in usa at what time\"",
"\"Typhoon Soulik in New York causing widespread flooding\"",
"typhoon -usa",
"\"New York in the path of typhoon Soulik\"",
"Typhoon Soulik storm surge, USA",
"\"New York during typhoon Soulik\"",
"\"Typhoon Sou... | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%83%9C%ED%92%8D_%EC%86%94%EB%A6%AD_(2018%EB%85%84)",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nSince \"Typhoon Soulik\" is a specific typhoon, it makes sense to extract events related to the storm itself or its impact. The query does not mention any human-related events or any other context, which suggests that extracting the typhoon itself and its effects would be a good idea.\n\nEVENTS:\n1. The storm hits a coastal area\n2. The storm causes widespread flooding\n3. A massive storm surge is generated by the typhoon\n4. Thousands of people are affected by the storm\n5. The storm causes widespread damage to infrastructure",
"en_article_excerpt": "Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.",
"en_article_url": "https://en.wikipedia.org/wiki/Typhoon_Soulik_(2018)",
"event_category": "disasters",
"event_name": "typhoon_soulik",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThe given query is a search query for Typhoon Soulik. It's likely that it could be visualized in a video as a destructive event, possibly a devastating storm with severe damage, given the description. Considering this, we should focus on events related to severe weather, destruction, and disaster.\n\nEVENTS:\n1. Severe weather\n2. Destruction\n3. Disaster\n4. Natural disaster",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nThe query seems to be asking for events that are likely to be associated with Typhoon Soulik, which made landfall in Korea. Given the mention of \"the deadliest typhoon to strike the Korean Peninsula\" and \"since Khanun in 2012\", it is reasonable to assume that Typhoon Soulik is likely to be the event in question. Therefore, the event that makes sense to extract is the typhoon itself.\n\nEVENTS:\n1. Typhoon Soulik\n2. Strike\n3. Landfall\n4. Strike\n5. Landfall",
"video_URL": "http://twitter.com/i/status/1032540315515682816",
"video_description": "#태풍 #솔릭 예상 경로! 동쪽으로 방향 틀어",
"video_id": "1032540315515682816"
} | 16 | [
"The image is a weather map showing the geographical regions of North Korea and South Korea, along with parts of China, Japan, and Russia. The map displays temperature data, with various cities marked and their respective temperatures noted in degrees Celsius. Two large, colorful spirals are depicted over the ocean... | 1032540315515682816 | The video shows a continuous display of North Korea and South Korea with temperature data displayed on a weather map. The map includes large, colorful spirals near South Korea and Japan, indicating the presence of cyclones or typhoons. The map source is "windy.com." The video repeats this process, showing temperature data and cyclone/typhoon indicators across the region, with cities and their temperatures marked. |
Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012. | [
"Typhoon Soulik severe weather New York",
"Typhoon Soulik, USA, not available",
"typhoon Soulik New York",
"typhoon soulik usa not available time",
"\"Typhoon Soulik New York\"",
"\"Typhoon Soulik in the USA during NOT available time\"",
"Typhoon Soulik strikes New York, potentially causing widespread d... | [
"\"Typhoon Soulik in New York around [time period]\"",
"typhoon soulik southeast usa at this time",
"\"New York typhoon Soulik strikes\"",
"\"Typhoon Soulik in the USA for [insert time frame, e.g., '2020-2021']\"",
"\"Typhoon affecting New York in September\"",
"\"typhoon landfall in usa\"",
"\"Typhoon ... | [
"New York in Typhoon Soulik",
"\"typhoon soulik hits coastal areas in usa at what time\"",
"\"Typhoon Soulik in New York causing widespread flooding\"",
"typhoon -usa",
"\"New York in the path of typhoon Soulik\"",
"Typhoon Soulik storm surge, USA",
"\"New York during typhoon Soulik\"",
"\"Typhoon Sou... | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%83%9C%ED%92%8D_%EC%86%94%EB%A6%AD_(2018%EB%85%84)",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nSince \"Typhoon Soulik\" is a specific typhoon, it makes sense to extract events related to the storm itself or its impact. The query does not mention any human-related events or any other context, which suggests that extracting the typhoon itself and its effects would be a good idea.\n\nEVENTS:\n1. The storm hits a coastal area\n2. The storm causes widespread flooding\n3. A massive storm surge is generated by the typhoon\n4. Thousands of people are affected by the storm\n5. The storm causes widespread damage to infrastructure",
"en_article_excerpt": "Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.",
"en_article_url": "https://en.wikipedia.org/wiki/Typhoon_Soulik_(2018)",
"event_category": "disasters",
"event_name": "typhoon_soulik",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThe given query is a search query for Typhoon Soulik. It's likely that it could be visualized in a video as a destructive event, possibly a devastating storm with severe damage, given the description. Considering this, we should focus on events related to severe weather, destruction, and disaster.\n\nEVENTS:\n1. Severe weather\n2. Destruction\n3. Disaster\n4. Natural disaster",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nThe query seems to be asking for events that are likely to be associated with Typhoon Soulik, which made landfall in Korea. Given the mention of \"the deadliest typhoon to strike the Korean Peninsula\" and \"since Khanun in 2012\", it is reasonable to assume that Typhoon Soulik is likely to be the event in question. Therefore, the event that makes sense to extract is the typhoon itself.\n\nEVENTS:\n1. Typhoon Soulik\n2. Strike\n3. Landfall\n4. Strike\n5. Landfall",
"video_URL": "http://twitter.com/i/status/1032471495958618112",
"video_description": "어제 저녁부터 비바람이 거세지다 잠잠하다 하더니, 지금은 세차게 몰아치는 중입니다 ㅠㅠ #태풍솔릭 #제주도 #제주시",
"video_id": "1032471495958618112"
} | 16 | [
"The image shows a dense cluster of trees and foliage. There is no clear depiction of any specific event or activity; it primarily features a natural, wooded environment.",
"The current frame continues to display a dense cluster of trees and foliage, maintaining the natural, wooded environment from the previous f... | 1032471495958618112 | A serene wooded environment with dense foliage and a gentle breeze continues throughout the video, showcasing the tranquility and peaceful atmosphere. The scene is set in a natural setting, devoid of any specific events or activities, allowing the viewers to immerse themselves in the calm environment. |
Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012. | [
"Typhoon Soulik severe weather New York",
"Typhoon Soulik, USA, not available",
"typhoon Soulik New York",
"typhoon soulik usa not available time",
"\"Typhoon Soulik New York\"",
"\"Typhoon Soulik in the USA during NOT available time\"",
"Typhoon Soulik strikes New York, potentially causing widespread d... | [
"\"Typhoon Soulik in New York around [time period]\"",
"typhoon soulik southeast usa at this time",
"\"New York typhoon Soulik strikes\"",
"\"Typhoon Soulik in the USA for [insert time frame, e.g., '2020-2021']\"",
"\"Typhoon affecting New York in September\"",
"\"typhoon landfall in usa\"",
"\"Typhoon ... | [
"New York in Typhoon Soulik",
"\"typhoon soulik hits coastal areas in usa at what time\"",
"\"Typhoon Soulik in New York causing widespread flooding\"",
"typhoon -usa",
"\"New York in the path of typhoon Soulik\"",
"Typhoon Soulik storm surge, USA",
"\"New York during typhoon Soulik\"",
"\"Typhoon Sou... | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%83%9C%ED%92%8D_%EC%86%94%EB%A6%AD_(2018%EB%85%84)",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nSince \"Typhoon Soulik\" is a specific typhoon, it makes sense to extract events related to the storm itself or its impact. The query does not mention any human-related events or any other context, which suggests that extracting the typhoon itself and its effects would be a good idea.\n\nEVENTS:\n1. The storm hits a coastal area\n2. The storm causes widespread flooding\n3. A massive storm surge is generated by the typhoon\n4. Thousands of people are affected by the storm\n5. The storm causes widespread damage to infrastructure",
"en_article_excerpt": "Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.",
"en_article_url": "https://en.wikipedia.org/wiki/Typhoon_Soulik_(2018)",
"event_category": "disasters",
"event_name": "typhoon_soulik",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThe given query is a search query for Typhoon Soulik. It's likely that it could be visualized in a video as a destructive event, possibly a devastating storm with severe damage, given the description. Considering this, we should focus on events related to severe weather, destruction, and disaster.\n\nEVENTS:\n1. Severe weather\n2. Destruction\n3. Disaster\n4. Natural disaster",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nThe query seems to be asking for events that are likely to be associated with Typhoon Soulik, which made landfall in Korea. Given the mention of \"the deadliest typhoon to strike the Korean Peninsula\" and \"since Khanun in 2012\", it is reasonable to assume that Typhoon Soulik is likely to be the event in question. Therefore, the event that makes sense to extract is the typhoon itself.\n\nEVENTS:\n1. Typhoon Soulik\n2. Strike\n3. Landfall\n4. Strike\n5. Landfall",
"video_URL": "http://twitter.com/i/status/1032234901897015298",
"video_description": "일본 가고시마에 상륙한 태풍 솔릭 위력ㅠㅠㅠ 8월19일자",
"video_id": "1032234901897015298"
} | 16 | [
"The image shows a strong storm with heavy rain and strong winds, affecting the area. There are cars visible on the street, and the visibility is low due to the weather conditions. The text indicates that this is in Amami Ōshima, with a severe typhoon warning for Typhoon 19. The time is noted as before 6 PM.",
"T... | 1032234901897015298 | A severe typhoon is depicted in the Amami Ōshima area, with heavy rain, strong winds, and low visibility due to the storm. The storm has a strong impact on the region, with two cars on the street struggling against the adverse weather. The typhoon warning for Typhoon 19 is highlighted, emphasizing its severity and the need for caution. The storm is expected to continue its impact until it passes. |
Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012. | [
"Typhoon Soulik severe weather New York",
"Typhoon Soulik, USA, not available",
"typhoon Soulik New York",
"typhoon soulik usa not available time",
"\"Typhoon Soulik New York\"",
"\"Typhoon Soulik in the USA during NOT available time\"",
"Typhoon Soulik strikes New York, potentially causing widespread d... | [
"\"Typhoon Soulik in New York around [time period]\"",
"typhoon soulik southeast usa at this time",
"\"New York typhoon Soulik strikes\"",
"\"Typhoon Soulik in the USA for [insert time frame, e.g., '2020-2021']\"",
"\"Typhoon affecting New York in September\"",
"\"typhoon landfall in usa\"",
"\"Typhoon ... | [
"New York in Typhoon Soulik",
"\"typhoon soulik hits coastal areas in usa at what time\"",
"\"Typhoon Soulik in New York causing widespread flooding\"",
"typhoon -usa",
"\"New York in the path of typhoon Soulik\"",
"Typhoon Soulik storm surge, USA",
"\"New York during typhoon Soulik\"",
"\"Typhoon Sou... | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%83%9C%ED%92%8D_%EC%86%94%EB%A6%AD_(2018%EB%85%84)",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nSince \"Typhoon Soulik\" is a specific typhoon, it makes sense to extract events related to the storm itself or its impact. The query does not mention any human-related events or any other context, which suggests that extracting the typhoon itself and its effects would be a good idea.\n\nEVENTS:\n1. The storm hits a coastal area\n2. The storm causes widespread flooding\n3. A massive storm surge is generated by the typhoon\n4. Thousands of people are affected by the storm\n5. The storm causes widespread damage to infrastructure",
"en_article_excerpt": "Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.",
"en_article_url": "https://en.wikipedia.org/wiki/Typhoon_Soulik_(2018)",
"event_category": "disasters",
"event_name": "typhoon_soulik",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThe given query is a search query for Typhoon Soulik. It's likely that it could be visualized in a video as a destructive event, possibly a devastating storm with severe damage, given the description. Considering this, we should focus on events related to severe weather, destruction, and disaster.\n\nEVENTS:\n1. Severe weather\n2. Destruction\n3. Disaster\n4. Natural disaster",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nThe query seems to be asking for events that are likely to be associated with Typhoon Soulik, which made landfall in Korea. Given the mention of \"the deadliest typhoon to strike the Korean Peninsula\" and \"since Khanun in 2012\", it is reasonable to assume that Typhoon Soulik is likely to be the event in question. Therefore, the event that makes sense to extract is the typhoon itself.\n\nEVENTS:\n1. Typhoon Soulik\n2. Strike\n3. Landfall\n4. Strike\n5. Landfall",
"video_URL": "http://youtube.com/watch?v=_V6HiUPBShQ",
"video_description": "태풍 피해 전해드렸는데요, 이렇게 제주와 일부 내륙에 피해가 집중된 건 예보와 달리 솔릭의 위력이 급격히 떨어졌기 때문입니다. 시속 100km를 넘나드는 강풍과 비를 동반해 모두를 긴장시켰던 태풍이 왜 이렇게 빠르게 약해졌는지, 엄민재 기자가 전해드립니다. 뉴스 원문 기사 보기 https://news.sbs.co.kr/news/endPage.do?news_id=N1004904932 [태풍 '솔릭' 북상] 기사 더보기 https://news.sbs.co.kr/news/newsHotIssueList.do?tagId=10000046975 #SBS뉴스 #SBSNEWS #SBS_NEWS #에스비에스뉴스 #태풍_'솔릭'_북상 #8뉴스 ▶SBS 뉴스 채널 구독하기 : https://n.sbs.co.kr/youtube ▶SBS 뉴스 라이브 : https://n.sbs.co.kr/youtubeLive , https://n.sbs.co.kr/live ▶대한민국 뉴스리더 SBS◀ 홈페이지: https://n.sbs.co.kr 페이스북: https://www.facebook.com/sbs8news 트위터: https://www.twitter.com/sbs8news",
"video_id": "_V6HiUPBShQ"
} | 16 | [
"A news anchor is delivering a weather report, discussing a typhoon moving rapidly. The background shows a map with weather patterns indicating the typhoon's path.",
"The news anchor continues the weather report, emphasizing the rapid movement of the typhoon. The background map displays the typhoon's path and its... | _V6HiUPBShQ | A typhoon named Solik is approaching Japan. The typhoon's rapid movement is the primary focus, with the typhoon's trajectory and speed being discussed in detail. The typhoon's impact on Japan is severe, including heavy rainfall, strong winds, and potential damage to infrastructure and ecosystems. Typhoon Solik is making landfall in Japan, bringing its strong winds and heavy rainfall, posing an immediate threat to maritime activities. |
Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012. | [
"Typhoon Soulik severe weather New York",
"Typhoon Soulik, USA, not available",
"typhoon Soulik New York",
"typhoon soulik usa not available time",
"\"Typhoon Soulik New York\"",
"\"Typhoon Soulik in the USA during NOT available time\"",
"Typhoon Soulik strikes New York, potentially causing widespread d... | [
"\"Typhoon Soulik in New York around [time period]\"",
"typhoon soulik southeast usa at this time",
"\"New York typhoon Soulik strikes\"",
"\"Typhoon Soulik in the USA for [insert time frame, e.g., '2020-2021']\"",
"\"Typhoon affecting New York in September\"",
"\"typhoon landfall in usa\"",
"\"Typhoon ... | [
"New York in Typhoon Soulik",
"\"typhoon soulik hits coastal areas in usa at what time\"",
"\"Typhoon Soulik in New York causing widespread flooding\"",
"typhoon -usa",
"\"New York in the path of typhoon Soulik\"",
"Typhoon Soulik storm surge, USA",
"\"New York during typhoon Soulik\"",
"\"Typhoon Sou... | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%83%9C%ED%92%8D_%EC%86%94%EB%A6%AD_(2018%EB%85%84)",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nSince \"Typhoon Soulik\" is a specific typhoon, it makes sense to extract events related to the storm itself or its impact. The query does not mention any human-related events or any other context, which suggests that extracting the typhoon itself and its effects would be a good idea.\n\nEVENTS:\n1. The storm hits a coastal area\n2. The storm causes widespread flooding\n3. A massive storm surge is generated by the typhoon\n4. Thousands of people are affected by the storm\n5. The storm causes widespread damage to infrastructure",
"en_article_excerpt": "Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.",
"en_article_url": "https://en.wikipedia.org/wiki/Typhoon_Soulik_(2018)",
"event_category": "disasters",
"event_name": "typhoon_soulik",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThe given query is a search query for Typhoon Soulik. It's likely that it could be visualized in a video as a destructive event, possibly a devastating storm with severe damage, given the description. Considering this, we should focus on events related to severe weather, destruction, and disaster.\n\nEVENTS:\n1. Severe weather\n2. Destruction\n3. Disaster\n4. Natural disaster",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nThe query seems to be asking for events that are likely to be associated with Typhoon Soulik, which made landfall in Korea. Given the mention of \"the deadliest typhoon to strike the Korean Peninsula\" and \"since Khanun in 2012\", it is reasonable to assume that Typhoon Soulik is likely to be the event in question. Therefore, the event that makes sense to extract is the typhoon itself.\n\nEVENTS:\n1. Typhoon Soulik\n2. Strike\n3. Landfall\n4. Strike\n5. Landfall",
"video_URL": "http://youtube.com/watch?v=Aiq8I97bscw",
"video_description": "제19호 태풍 '솔릭'이 일본 남부 해상에서 한반도를 향해 북진하고 있습니다. '솔릭'은 강한 중형급 태풍으로 중심기압은 960hPa, 강풍 반경은 340㎞이라고 하는데요. 만일 태풍이 우리나라를 관통하면 2012년 9월 '산바' 이후 약 6년 만으로 큰 피해가 우려됩니다. 자세한 내용 영상으로 보시죠. ◆ 연합뉴스 홈페이지→ http://www.yna.co.kr/ ◆ 이 시각 많이 본 기사 → https://goo.gl/VQTsSZ ◆ 오늘의 핫뉴스 → https://goo.gl/WyGXpG ◆ 현장영상 → https://goo.gl/5aZcx8 ◆ 카드뉴스 →https://goo.gl/QKfDTH ◆연합뉴스 공식 SNS◆ ◇페이스북→ https://www.facebook.com/yonhap/ ▣ 연합뉴스 유튜브 채널 구독 : https://goo.gl/pL7TmT",
"video_id": "Aiq8I97bscw"
} | 16 | [
"The frame shows a turbulent ocean with large waves crashing against rocky outcrops. The Korean text indicates that Typhoon \"Sulak\" is affecting the northern coastal areas of Japan, causing heavy rainfall.",
"The frame depicts a weather map illustrating the path and intensity of Typhoon \"Sulak.\" The map highl... | Aiq8I97bscw | A severe typhoon, known as Typhoon "Sulak," affects the northern coastal areas of Japan, causing heavy rainfall and strong winds. The typhoon has a high pressure of 960hPa and a wind force radius of 340 km. The storm intensifies over the East China Sea, leading to widespread damage and flooding along the coast. Severe weather conditions persist for several days, with reports of torrential rain, strong winds, and massive waves. The typhoon's projected path and impact on the entire country are being closely monitored, with a possibility of further damage and landslides. The public is urged to be prepared and take necessary precautions to avoid damage. |
Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012. | [
"Typhoon Soulik severe weather New York",
"Typhoon Soulik, USA, not available",
"typhoon Soulik New York",
"typhoon soulik usa not available time",
"\"Typhoon Soulik New York\"",
"\"Typhoon Soulik in the USA during NOT available time\"",
"Typhoon Soulik strikes New York, potentially causing widespread d... | [
"\"Typhoon Soulik in New York around [time period]\"",
"typhoon soulik southeast usa at this time",
"\"New York typhoon Soulik strikes\"",
"\"Typhoon Soulik in the USA for [insert time frame, e.g., '2020-2021']\"",
"\"Typhoon affecting New York in September\"",
"\"typhoon landfall in usa\"",
"\"Typhoon ... | [
"New York in Typhoon Soulik",
"\"typhoon soulik hits coastal areas in usa at what time\"",
"\"Typhoon Soulik in New York causing widespread flooding\"",
"typhoon -usa",
"\"New York in the path of typhoon Soulik\"",
"Typhoon Soulik storm surge, USA",
"\"New York during typhoon Soulik\"",
"\"Typhoon Sou... | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%83%9C%ED%92%8D_%EC%86%94%EB%A6%AD_(2018%EB%85%84)",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nSince \"Typhoon Soulik\" is a specific typhoon, it makes sense to extract events related to the storm itself or its impact. The query does not mention any human-related events or any other context, which suggests that extracting the typhoon itself and its effects would be a good idea.\n\nEVENTS:\n1. The storm hits a coastal area\n2. The storm causes widespread flooding\n3. A massive storm surge is generated by the typhoon\n4. Thousands of people are affected by the storm\n5. The storm causes widespread damage to infrastructure",
"en_article_excerpt": "Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.",
"en_article_url": "https://en.wikipedia.org/wiki/Typhoon_Soulik_(2018)",
"event_category": "disasters",
"event_name": "typhoon_soulik",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThe given query is a search query for Typhoon Soulik. It's likely that it could be visualized in a video as a destructive event, possibly a devastating storm with severe damage, given the description. Considering this, we should focus on events related to severe weather, destruction, and disaster.\n\nEVENTS:\n1. Severe weather\n2. Destruction\n3. Disaster\n4. Natural disaster",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nThe query seems to be asking for events that are likely to be associated with Typhoon Soulik, which made landfall in Korea. Given the mention of \"the deadliest typhoon to strike the Korean Peninsula\" and \"since Khanun in 2012\", it is reasonable to assume that Typhoon Soulik is likely to be the event in question. Therefore, the event that makes sense to extract is the typhoon itself.\n\nEVENTS:\n1. Typhoon Soulik\n2. Strike\n3. Landfall\n4. Strike\n5. Landfall",
"video_URL": "http://youtube.com/watch?v=25LVZnXdx0A",
"video_description": "중부지방을 중심으로 연일 매섭게 비가 쏟아지고 있다. 긴급상황에 대비해야 할 소방서까지 침수된 모습이 SNS를 통해 전파되면서 불안이 가중되고 있다. 비도 이 정도면 재해 수준. 아무 일도 없다는 듯 태풍 솔릭은 지나갔지만, 오히려 비의 공포는 더해져만 가는데... 제작 : 박태호PD(ptho@ytnplus.co.kr) ▶ 기사 원문 : http://www.ytn.co.kr/_ln/0103_201808292100061801 ▶ 제보 안내 : http://goo.gl/gEvsAL, 모바일앱, 8585@ytn.co.kr, #2424 ▣ YTN 유튜브 채널 구독 : http://goo.gl/Ytb5SZ [ 한국 뉴스 채널 와이티엔 / Korea News Channel YTN ]",
"video_id": "25LVZnXdx0A"
} | 16 | [
"A vehicle is navigating through a flooded street at night. The headlights illuminate the water, and there are no people visible. The Korean text on the screen reads \"전국이 폭우로 몸살을 앓고 있다,\" which translates to \"The whole country is suffering from torrential rain.\"",
"A close-up of a person's mouth is visible, li... | 25LVZnXdx0A | A severe flood in South Korea caused by typhoon 'Sokoli' left the country suffering widespread damage and disruption, particularly in the central region. A broadcasting channel, YTN, reported on the situation, with footage showing the impact of the flood on infrastructure, including a fire station and a flooded area. The floodwaters overwhelmed critical barriers, causing significant damage and displacement of people. However, there were also reports of people standing on rooftops, conducting repairs, and expressing frustration with the situation. The flood also caused widespread panic and disruption to daily life. |
Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012. | [
"Typhoon Soulik severe weather New York",
"Typhoon Soulik, USA, not available",
"typhoon Soulik New York",
"typhoon soulik usa not available time",
"\"Typhoon Soulik New York\"",
"\"Typhoon Soulik in the USA during NOT available time\"",
"Typhoon Soulik strikes New York, potentially causing widespread d... | [
"\"Typhoon Soulik in New York around [time period]\"",
"typhoon soulik southeast usa at this time",
"\"New York typhoon Soulik strikes\"",
"\"Typhoon Soulik in the USA for [insert time frame, e.g., '2020-2021']\"",
"\"Typhoon affecting New York in September\"",
"\"typhoon landfall in usa\"",
"\"Typhoon ... | [
"New York in Typhoon Soulik",
"\"typhoon soulik hits coastal areas in usa at what time\"",
"\"Typhoon Soulik in New York causing widespread flooding\"",
"typhoon -usa",
"\"New York in the path of typhoon Soulik\"",
"Typhoon Soulik storm surge, USA",
"\"New York during typhoon Soulik\"",
"\"Typhoon Sou... | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%83%9C%ED%92%8D_%EC%86%94%EB%A6%AD_(2018%EB%85%84)",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nSince \"Typhoon Soulik\" is a specific typhoon, it makes sense to extract events related to the storm itself or its impact. The query does not mention any human-related events or any other context, which suggests that extracting the typhoon itself and its effects would be a good idea.\n\nEVENTS:\n1. The storm hits a coastal area\n2. The storm causes widespread flooding\n3. A massive storm surge is generated by the typhoon\n4. Thousands of people are affected by the storm\n5. The storm causes widespread damage to infrastructure",
"en_article_excerpt": "Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.",
"en_article_url": "https://en.wikipedia.org/wiki/Typhoon_Soulik_(2018)",
"event_category": "disasters",
"event_name": "typhoon_soulik",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nThe given query is a search query for Typhoon Soulik. It's likely that it could be visualized in a video as a destructive event, possibly a devastating storm with severe damage, given the description. Considering this, we should focus on events related to severe weather, destruction, and disaster.\n\nEVENTS:\n1. Severe weather\n2. Destruction\n3. Disaster\n4. Natural disaster",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Typhoon Soulik was an unusually large, and the deadliest typhoon to strike the Korean Peninsula since Khanun in 2012.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nThe query seems to be asking for events that are likely to be associated with Typhoon Soulik, which made landfall in Korea. Given the mention of \"the deadliest typhoon to strike the Korean Peninsula\" and \"since Khanun in 2012\", it is reasonable to assume that Typhoon Soulik is likely to be the event in question. Therefore, the event that makes sense to extract is the typhoon itself.\n\nEVENTS:\n1. Typhoon Soulik\n2. Strike\n3. Landfall\n4. Strike\n5. Landfall",
"video_URL": "http://youtube.com/watch?v=BRxOTPDR0g4",
"video_description": "건물 공사 구조물이 무너지고 교회 첨탑과 호텔 지붕도 강풍에 날려 추락했습니다. 아름드리 가로수는 뿌리째 뽑혔습니다. 지난 2010년 태풍 곤파스가 몰고 온 강풍의 위력입니다. 전남 홍도에 초속 42.9m의 폭풍이 불었고 태안 38.1m, 군산 33.6m, 서울도 21.6m의 강풍이 기록됐습니다. 바람 태풍 곤파스로 6명이 숨지고 11명이 부상했으며, 1,673억 원의 재산 피해가 발생했습니다. 북상하는 19호 태풍 '솔릭'은 8년 전 '곤파스'와 비슷한 경로로 움직이고 있습니다. 내일 새벽, 제주도 서쪽을 거쳐 서해안을 따라 북상한 뒤 태안반도에 상륙해 수도권을 통과할 것으로 예상됩니다. 한반도 주변 수온이 28도 정도로 무척 높아, 곤파스보다 강한 세력을 유지한 채 북상할 가능성이 큽니다. 기상청은 태풍 '솔릭'으로 인해 목요일과 금요일 사이 전국에 최고 400mm의 폭우가 쏟아지고 초속 40m가 넘는 강풍이 불 것으로 전망했습니다. 취재기자 : 정혜윤 영상편집 : 연진영 자막뉴스 제작 : 육지혜 에디터 ▶ 기사 원문 : http://www.ytn.co.kr/_ln/0108_201808220940500446 ▶ 제보 안내 : http://goo.gl/gEvsAL, 모바일앱, 8585@ytn.co.kr, #2424 ▣ YTN 유튜브 채널 구독 : http://goo.gl/Ytb5SZ [ 한국 뉴스 채널 와이티엔 / Korea News Channel YTN ]",
"video_id": "BRxOTPDR0g4"
} | 16 | [
"A large metal structure, possibly a bus stop or shelter, has collapsed onto a busy city street, creating a disruption. Several people are seen walking nearby, with some heading towards the collapsed structure. The background shows multiple vehicles, including a police car and buses, indicating this is a major inte... | BRxOTPDR0g4 | A severe typhoon, Typhoon Kong-phasa, had a significant impact on the city of Seoul and surrounding areas in South Korea in 2010. The typhoon caused widespread damage to buildings, infrastructure, and agricultural facilities. Emergency services were deployed to assist with the relief efforts, and the number of casualties and damage was substantial.
A strong storm, Typhoon Solik, approached the region, bringing similar characteristics to Kong-phasa, and was expected to impact the same regions. The typhoon's strong winds caused significant damage to trees and buildings, with the path of the storm visible on a map provided. The storm's path was shifted westward towards the Korean Peninsula.
The typhoon resulted in an estimated 6 deaths and 11 injuries, with significant damage to agricultural facilities, and resulted in 1.673 billion won in property damage. The typhoon also caused heavy rain and strong winds, causing widespread disruptions to daily life. |
The 2021 East Asia sandstorm was a meteorological phenomenon that began in the Eastern Gobi desert steppe on March 14, and subsequently spread to the entire Mongolian Plateau South, the Loess Plateau, the North China Plain and the Korean Peninsula. | [
"\"2021 East Asian sandstorm in Tokyo\"",
"\"Sandstorms affecting East Asia during [insert time] with relevant event details\"",
"Sandstorms in East Asia",
"\"Sandstorms in 2021 East Asia",
"\"east asian dust storms\"",
"Dust storms affecting regions in East Asia",
"\"Sandstorms in the Eastern Gobi Dese... | [
"\"2021 East Asia sandstorm in the Eastern Gobi desert\"",
"\"Sandstorms in East Asia, particularly affecting the Eastern Gobi desert steppe, in an unspecified time frame, near the region\"",
"The East Asia region is experiencing a severe sandstorm.",
"\"The extreme weather event in the Eastern Gobi Desert, b... | [
"Sandstorm - East Asia - East Asia",
"east asia dust storms",
"dust storms in East Asia, occurring between 2020 and 2021, affecting regions surrounding the region.",
"\"2021 East Asia sandstorm\"",
"\"Global dust storms affecting East Asia, especially during periods of strong winds\"",
"\"Strong winds, du... | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%99%A9%EC%82%AC",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2021 East Asia sandstorm was a meteorological phenomenon that began in the Eastern Gobi desert steppe on March 14, and subsequently spread to the entire Mongolian Plateau South, the Loess Plateau, the North China Plain and the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\n\nIt seems logical to extract events that are related to weather phenomena. Given the query's mention of \"sandstorm\", \"meteorological phenomenon\", and \"weather\" events, it is reasonable to extract weather-related events that could be visualized in a video, such as sandstorms, windstorms, or other weather-related activities.\n\nEVENTS:\n\n1. Sandstorm\n2. Windstorm\n3. Storms\n4. Weather\n5. Cyclone",
"en_article_excerpt": "The 2021 East Asia sandstorm was a meteorological phenomenon that began in the Eastern Gobi desert steppe on March 14, and subsequently spread to the entire Mongolian Plateau South, the Loess Plateau, the North China Plain and the Korean Peninsula.",
"en_article_url": "https://en.wikipedia.org/wiki/2021_East_Asia_sandstorm",
"event_category": "disasters",
"event_name": "2021_sandstorm",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2021 East Asia sandstorm was a meteorological phenomenon that began in the Eastern Gobi desert steppe on March 14, and subsequently spread to the entire Mongolian Plateau South, the Loess Plateau, the North China Plain and the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nIt is not necessary to extract events in this query as it seems to be describing a meteorological event, rather than a video event. Therefore, we do not need to consider visual events or determine their relevance to a video.\n\nEVENTS:\n1. Sandstorms\n2. Eastern Gobi desert steppe\n3. Mongolian Plateau South\n4. Loess Plateau\n5. North China Plain\n6. Korean Peninsula",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2021 East Asia sandstorm was a meteorological phenomenon that began in the Eastern Gobi desert steppe on March 14, and subsequently spread to the entire Mongolian Plateau South, the Loess Plateau, the North China Plain and the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nThe 2021 East Asia sandstorm is a meteorological phenomenon, so it makes sense to extract events that could be visualized in a video of people experiencing or affected by the sandstorm. The word \"sandstorm\" itself is a fairly vivid description, making it likely to be seen in a video. Also, the geographical locations mentioned (Eastern Gobi desert steppe, Mongolian Plateau South, Loess Plateau, North China Plain, and Korean Peninsula) suggest that people experiencing the sandstorm might be captured in a video, especially if the video is about natural disasters or extreme weather events.\n\nEVENTS:\n1. A sandstorm breaking out in the Eastern Gobi desert steppe\n2. A person being affected by the sandstorm\n3. The spread of the sandstorm to the entire Mongolian Plateau\n4. People taking shelter from the sandstorm in a nearby location\n5. A group of people seeking medical attention due to sandstorm-related injuries",
"video_URL": "http://youtube.com/watch?v=Yddr-IKR_KA",
"video_description": "중국 둔황에 초대형 모래폭풍이 불어닥치면서눈 깜짝할 사이에 도시 전체가 암흑으로 뒤덮었습니다. [Ch.19] 사실을 보고 진실을 말합니다. 👍🏻 공식 홈페이지 http://news.tvchosun.com/ 👍🏻 공식 페이스북 https://www.facebook.com/tvchosunnews/ 👍🏻 공식 트위터 https://twitter.com/TVChosunNews * 뉴스제보 : 이메일(tvchosun@chosun.com), 카카오톡(tv조선제보), 전화(1661-0190)",
"video_id": "Yddr-IKR_KA"
} | 16 | [
"A news anchor is seated in a studio, wearing a suit and holding papers, presumably delivering a news broadcast. The background features a large number \"24\" and some Korean text, indicating the time or channel name.",
"The news anchor is seated in the studio, wearing a suit and speaking, likely continuing the n... | Yddr-IKR_KA | A massive dust storm has engulfed a rural and urban area, with significant visibility reduced and hazardous conditions reported. The storm's impact is severe, covering both rural and urban areas with a thick, yellowish haze, and reducing visibility to near zero. Residents have taken shelter indoors, and people have also taken precautions to avoid the storm. The storm's height is approximately 100 meters, and the dust cloud has spread across the landscape. The scene continues to show the full extent of the storm's effects, with no visible human activity and no visible structures or vegetation. The storm's intensity has become increasingly severe, with reports of hazardous conditions and precautions being taken to mitigate its effects. |
The 2021 East Asia sandstorm was a meteorological phenomenon that began in the Eastern Gobi desert steppe on March 14, and subsequently spread to the entire Mongolian Plateau South, the Loess Plateau, the North China Plain and the Korean Peninsula. | [
"\"2021 East Asian sandstorm in Tokyo\"",
"\"Sandstorms affecting East Asia during [insert time] with relevant event details\"",
"Sandstorms in East Asia",
"\"Sandstorms in 2021 East Asia",
"\"east asian dust storms\"",
"Dust storms affecting regions in East Asia",
"\"Sandstorms in the Eastern Gobi Dese... | [
"\"2021 East Asia sandstorm in the Eastern Gobi desert\"",
"\"Sandstorms in East Asia, particularly affecting the Eastern Gobi desert steppe, in an unspecified time frame, near the region\"",
"The East Asia region is experiencing a severe sandstorm.",
"\"The extreme weather event in the Eastern Gobi Desert, b... | [
"Sandstorm - East Asia - East Asia",
"east asia dust storms",
"dust storms in East Asia, occurring between 2020 and 2021, affecting regions surrounding the region.",
"\"2021 East Asia sandstorm\"",
"\"Global dust storms affecting East Asia, especially during periods of strong winds\"",
"\"Strong winds, du... | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%99%A9%EC%82%AC",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2021 East Asia sandstorm was a meteorological phenomenon that began in the Eastern Gobi desert steppe on March 14, and subsequently spread to the entire Mongolian Plateau South, the Loess Plateau, the North China Plain and the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\n\nIt seems logical to extract events that are related to weather phenomena. Given the query's mention of \"sandstorm\", \"meteorological phenomenon\", and \"weather\" events, it is reasonable to extract weather-related events that could be visualized in a video, such as sandstorms, windstorms, or other weather-related activities.\n\nEVENTS:\n\n1. Sandstorm\n2. Windstorm\n3. Storms\n4. Weather\n5. Cyclone",
"en_article_excerpt": "The 2021 East Asia sandstorm was a meteorological phenomenon that began in the Eastern Gobi desert steppe on March 14, and subsequently spread to the entire Mongolian Plateau South, the Loess Plateau, the North China Plain and the Korean Peninsula.",
"en_article_url": "https://en.wikipedia.org/wiki/2021_East_Asia_sandstorm",
"event_category": "disasters",
"event_name": "2021_sandstorm",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2021 East Asia sandstorm was a meteorological phenomenon that began in the Eastern Gobi desert steppe on March 14, and subsequently spread to the entire Mongolian Plateau South, the Loess Plateau, the North China Plain and the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nIt is not necessary to extract events in this query as it seems to be describing a meteorological event, rather than a video event. Therefore, we do not need to consider visual events or determine their relevance to a video.\n\nEVENTS:\n1. Sandstorms\n2. Eastern Gobi desert steppe\n3. Mongolian Plateau South\n4. Loess Plateau\n5. North China Plain\n6. Korean Peninsula",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2021 East Asia sandstorm was a meteorological phenomenon that began in the Eastern Gobi desert steppe on March 14, and subsequently spread to the entire Mongolian Plateau South, the Loess Plateau, the North China Plain and the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nThe 2021 East Asia sandstorm is a meteorological phenomenon, so it makes sense to extract events that could be visualized in a video of people experiencing or affected by the sandstorm. The word \"sandstorm\" itself is a fairly vivid description, making it likely to be seen in a video. Also, the geographical locations mentioned (Eastern Gobi desert steppe, Mongolian Plateau South, Loess Plateau, North China Plain, and Korean Peninsula) suggest that people experiencing the sandstorm might be captured in a video, especially if the video is about natural disasters or extreme weather events.\n\nEVENTS:\n1. A sandstorm breaking out in the Eastern Gobi desert steppe\n2. A person being affected by the sandstorm\n3. The spread of the sandstorm to the entire Mongolian Plateau\n4. People taking shelter from the sandstorm in a nearby location\n5. A group of people seeking medical attention due to sandstorm-related injuries",
"video_URL": "http://youtube.com/watch?v=3BbP2OHdePY",
"video_description": "뿌연 하루였습니다. 몽골에서 시작된 모래폭풍이 중국을 거쳐 한반도에 상륙한 건데, 내일은 전국에서 황사가루가 섞인 '흙비'까지 내릴 전망입니다. [Ch.19] 사실을 보고 진실을 말합니다. 👍🏻 공식 홈페이지 http://news.tvchosun.com/ 👍🏻 공식 페이스북 https://www.facebook.com/tvchosunnews/ 👍🏻 공식 트위터 https://twitter.com/TVChosunNews * 뉴스제보 : 이메일(tvchosun@chosun.com), 카카오톡(tv조선제보), 전화(1661-0190)",
"video_id": "3BbP2OHdePY"
} | 16 | [
"A news anchor in a blue blazer stands in front of a large screen displaying a dust storm over a landscape. The screen shows a large cloud of dust, indicating a severe weather event. The text in Korean at the bottom of the screen suggests a news broadcast about a serious weather issue, with information about a dust... | 3BbP2OHdePY | The video depicts a severe dust storm affecting a cityscape, highlighting the impact on daily life, culture, and environment. Key points include:
* The dust storm is expected to continue or worsen the next day, with forecasts predicting a mix of dust, sand, and rain.
* The storm is expected to affect various regions of the country, including Seoul and its surroundings.
* The storm is expected to cause disruptions to daily life, including road closures, travel warnings, and cultural events.
* The storm may also affect the city's economy, with potential impacts on transportation, commerce, and tourism.
* The storm may also have an impact on the environment, with dust and sand affecting local wildlife and infrastructure.
* The storm is expected to affect various regions of the country, including the eastern and southern regions, with temperatures expected to be below average.
* The forecast for the next day is uncertain, but it is expected to be a mix of rain, dust, and sand.
* The storm may also have an impact on sports events, including a basketball game between Real Madrid and Chelsea. |
The 2021 East Asia sandstorm was a meteorological phenomenon that began in the Eastern Gobi desert steppe on March 14, and subsequently spread to the entire Mongolian Plateau South, the Loess Plateau, the North China Plain and the Korean Peninsula. | [
"\"2021 East Asian sandstorm in Tokyo\"",
"\"Sandstorms affecting East Asia during [insert time] with relevant event details\"",
"Sandstorms in East Asia",
"\"Sandstorms in 2021 East Asia",
"\"east asian dust storms\"",
"Dust storms affecting regions in East Asia",
"\"Sandstorms in the Eastern Gobi Dese... | [
"\"2021 East Asia sandstorm in the Eastern Gobi desert\"",
"\"Sandstorms in East Asia, particularly affecting the Eastern Gobi desert steppe, in an unspecified time frame, near the region\"",
"The East Asia region is experiencing a severe sandstorm.",
"\"The extreme weather event in the Eastern Gobi Desert, b... | [
"Sandstorm - East Asia - East Asia",
"east asia dust storms",
"dust storms in East Asia, occurring between 2020 and 2021, affecting regions surrounding the region.",
"\"2021 East Asia sandstorm\"",
"\"Global dust storms affecting East Asia, especially during periods of strong winds\"",
"\"Strong winds, du... | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%99%A9%EC%82%AC",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2021 East Asia sandstorm was a meteorological phenomenon that began in the Eastern Gobi desert steppe on March 14, and subsequently spread to the entire Mongolian Plateau South, the Loess Plateau, the North China Plain and the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\n\nIt seems logical to extract events that are related to weather phenomena. Given the query's mention of \"sandstorm\", \"meteorological phenomenon\", and \"weather\" events, it is reasonable to extract weather-related events that could be visualized in a video, such as sandstorms, windstorms, or other weather-related activities.\n\nEVENTS:\n\n1. Sandstorm\n2. Windstorm\n3. Storms\n4. Weather\n5. Cyclone",
"en_article_excerpt": "The 2021 East Asia sandstorm was a meteorological phenomenon that began in the Eastern Gobi desert steppe on March 14, and subsequently spread to the entire Mongolian Plateau South, the Loess Plateau, the North China Plain and the Korean Peninsula.",
"en_article_url": "https://en.wikipedia.org/wiki/2021_East_Asia_sandstorm",
"event_category": "disasters",
"event_name": "2021_sandstorm",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2021 East Asia sandstorm was a meteorological phenomenon that began in the Eastern Gobi desert steppe on March 14, and subsequently spread to the entire Mongolian Plateau South, the Loess Plateau, the North China Plain and the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nIt is not necessary to extract events in this query as it seems to be describing a meteorological event, rather than a video event. Therefore, we do not need to consider visual events or determine their relevance to a video.\n\nEVENTS:\n1. Sandstorms\n2. Eastern Gobi desert steppe\n3. Mongolian Plateau South\n4. Loess Plateau\n5. North China Plain\n6. Korean Peninsula",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2021 East Asia sandstorm was a meteorological phenomenon that began in the Eastern Gobi desert steppe on March 14, and subsequently spread to the entire Mongolian Plateau South, the Loess Plateau, the North China Plain and the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nThe 2021 East Asia sandstorm is a meteorological phenomenon, so it makes sense to extract events that could be visualized in a video of people experiencing or affected by the sandstorm. The word \"sandstorm\" itself is a fairly vivid description, making it likely to be seen in a video. Also, the geographical locations mentioned (Eastern Gobi desert steppe, Mongolian Plateau South, Loess Plateau, North China Plain, and Korean Peninsula) suggest that people experiencing the sandstorm might be captured in a video, especially if the video is about natural disasters or extreme weather events.\n\nEVENTS:\n1. A sandstorm breaking out in the Eastern Gobi desert steppe\n2. A person being affected by the sandstorm\n3. The spread of the sandstorm to the entire Mongolian Plateau\n4. People taking shelter from the sandstorm in a nearby location\n5. A group of people seeking medical attention due to sandstorm-related injuries",
"video_URL": "http://youtube.com/watch?v=6sLn3wHQvCQ",
"video_description": "초대형 모래폭풍에 산사태까지…재난영화 방불케한 순간 포착 (서울=연합뉴스) 중국 북서부 간쑤성에서 초대형 모래폭풍이 도시를 집어삼키고 산사태로 엄청난 흙더미가 쓰나미처럼 쏟아져 내리는 등 흡사 재난 영화의 한 장면 같은 순간들이 카메라에 포착됐습니다. 지난 25일 간쑤성 과저우현에서는 높이 수십 미터에 달하는 거대한 모래폭풍이 도시를 덮쳤습니다. 모래폭풍은 마치 거대한 모래 장벽 같은 모습이었습니다. 모래폭풍이 시야를 가리고 해를 가리면서 일대는 오렌지색으로 물들었습니다. 가시거리가 50m 미만, 공기 질이 최악 수준으로 내려앉으면서 대기오염 최고단계인 황색경보가 발령됐습니다. 한편 같은 날 간쑤성 룽난시에서는 산사태가 발생했습니다. 산 중턱의 한 부분에서부터 흙더미와 나무가 쓰나미처럼 산 아래로 쏟아져 내리면서 도로에 있던 관광객 수십 명이 비명을 지르며 도망쳤습니다. 이날 산사태로 차량 한 대가 파손됐지만 인명 피해는 없었다고 합니다. 영상으로 보시죠. 제작 : 김건태·안창주 영상 : 로이터 #연합뉴스 #중국 #황사 ◆ 연합뉴스 유튜브 : https://www.youtube.com/yonhap ◆ 연합뉴스 홈페이지→ http://www.yna.co.kr/ ◆ 연합뉴스 페이스북→ https://www.facebook.com/yonhap/ ◆ 연합뉴스 인스타 : https://goo.gl/UbqiQb ◆ 연합뉴스 비디오메타 채널https://www.youtube.com/channel/UCTMCrbtHU0M0SR6TuBrL4Pw",
"video_id": "6sLn3wHQvCQ"
} | 16 | [
"A large dust storm is approaching a city, obscuring the sky with a thick cloud of dust. The buildings in the foreground appear to be low-rise structures, and the sky is overcast with dust clouds. The text \"연합뉴스\" (Yonhap News) and \"실제상황\" (Actual Situation) are visible on the image, indicating that this is a new... | 6sLn3wHQvCQ | A severe sandstorm is depicted in a series of 16 consecutive frames, showcasing its devastating impact on a city. The storm begins as a dust cloud in the foreground, gradually intensifying and engulfing the city. The storm's severity is highlighted through various visual elements, such as the dense dust, towering clouds, and buildings obscured by the dust. As the storm persists, the environment becomes increasingly desolate, with trees bent, vegetation destroyed, and debris scattered. Emergency situations are depicted, including people fleeing and attempting to escape, while others are trying to help or assist. The storm eventually subsides, revealing a scene of destruction and devastation. |
The 2021 East Asia sandstorm was a meteorological phenomenon that began in the Eastern Gobi desert steppe on March 14, and subsequently spread to the entire Mongolian Plateau South, the Loess Plateau, the North China Plain and the Korean Peninsula. | [
"\"2021 East Asian sandstorm in Tokyo\"",
"\"Sandstorms affecting East Asia during [insert time] with relevant event details\"",
"Sandstorms in East Asia",
"\"Sandstorms in 2021 East Asia",
"\"east asian dust storms\"",
"Dust storms affecting regions in East Asia",
"\"Sandstorms in the Eastern Gobi Dese... | [
"\"2021 East Asia sandstorm in the Eastern Gobi desert\"",
"\"Sandstorms in East Asia, particularly affecting the Eastern Gobi desert steppe, in an unspecified time frame, near the region\"",
"The East Asia region is experiencing a severe sandstorm.",
"\"The extreme weather event in the Eastern Gobi Desert, b... | [
"Sandstorm - East Asia - East Asia",
"east asia dust storms",
"dust storms in East Asia, occurring between 2020 and 2021, affecting regions surrounding the region.",
"\"2021 East Asia sandstorm\"",
"\"Global dust storms affecting East Asia, especially during periods of strong winds\"",
"\"Strong winds, du... | {
"article_url": "https://ko.wikipedia.org/wiki/%ED%99%A9%EC%82%AC",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2021 East Asia sandstorm was a meteorological phenomenon that began in the Eastern Gobi desert steppe on March 14, and subsequently spread to the entire Mongolian Plateau South, the Loess Plateau, the North China Plain and the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\n\nIt seems logical to extract events that are related to weather phenomena. Given the query's mention of \"sandstorm\", \"meteorological phenomenon\", and \"weather\" events, it is reasonable to extract weather-related events that could be visualized in a video, such as sandstorms, windstorms, or other weather-related activities.\n\nEVENTS:\n\n1. Sandstorm\n2. Windstorm\n3. Storms\n4. Weather\n5. Cyclone",
"en_article_excerpt": "The 2021 East Asia sandstorm was a meteorological phenomenon that began in the Eastern Gobi desert steppe on March 14, and subsequently spread to the entire Mongolian Plateau South, the Loess Plateau, the North China Plain and the Korean Peninsula.",
"en_article_url": "https://en.wikipedia.org/wiki/2021_East_Asia_sandstorm",
"event_category": "disasters",
"event_name": "2021_sandstorm",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2021 East Asia sandstorm was a meteorological phenomenon that began in the Eastern Gobi desert steppe on March 14, and subsequently spread to the entire Mongolian Plateau South, the Loess Plateau, the North China Plain and the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nIt is not necessary to extract events in this query as it seems to be describing a meteorological event, rather than a video event. Therefore, we do not need to consider visual events or determine their relevance to a video.\n\nEVENTS:\n1. Sandstorms\n2. Eastern Gobi desert steppe\n3. Mongolian Plateau South\n4. Loess Plateau\n5. North China Plain\n6. Korean Peninsula",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: The 2021 East Asia sandstorm was a meteorological phenomenon that began in the Eastern Gobi desert steppe on March 14, and subsequently spread to the entire Mongolian Plateau South, the Loess Plateau, the North China Plain and the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nThe 2021 East Asia sandstorm is a meteorological phenomenon, so it makes sense to extract events that could be visualized in a video of people experiencing or affected by the sandstorm. The word \"sandstorm\" itself is a fairly vivid description, making it likely to be seen in a video. Also, the geographical locations mentioned (Eastern Gobi desert steppe, Mongolian Plateau South, Loess Plateau, North China Plain, and Korean Peninsula) suggest that people experiencing the sandstorm might be captured in a video, especially if the video is about natural disasters or extreme weather events.\n\nEVENTS:\n1. A sandstorm breaking out in the Eastern Gobi desert steppe\n2. A person being affected by the sandstorm\n3. The spread of the sandstorm to the entire Mongolian Plateau\n4. People taking shelter from the sandstorm in a nearby location\n5. A group of people seeking medical attention due to sandstorm-related injuries",
"video_URL": "http://youtube.com/watch?v=qaan1wMGnN0",
"video_description": "거대한 '모래 폭풍'이 하늘을 뒤덮었습니다. 지난 주말 몽골의 건조한 초원에서 발원한 '황사 바람'입니다. 태풍급 강풍을 동반한 모래 먼지가 한적한 작은 마을을 집어삼킬 듯한 기세입니다. 초원은 물론이고 주변 도시들도 앞을 분간하기 어려운 먼지 속에 갇혔습니다. '황색 폭풍'의 습격으로 몽골 초원지대 유목민 약 600명이 한때 실종됐습니다. 다행히 대부분 살아서 돌아왔지만 5살 어린 아이 를 포함해 10여 명은 결국 숨졌습니다. 최고 풍속 초속 30m가 넘는 모래 폭풍에 10층 높이의 철탑도 맥없이 부러졌습니다. 인근 지역 주민들은 모래 바람의 공포 속에, 정전 까지 돼 암흑과 추위에 떨었습니다. 몽골 초원과 가까운 중국 북쪽 네이멍구에서는 모래가 섞인 눈보라가 몰아쳤습니다. 강풍에 밀려온 모래와 눈이 민가를 뒤덮으면서 집안에 갇히는 주민들도 속출했습니다. 경찰 대여섯 명이 삽과 수레로 눈 모래를 걷어낸 지 2시간 만에 71살 할머니가 구조되기도 했습니다. 이번 황사는 강한 바람이 모래 먼지를 멀리까지 실어나르면서 '10년 만의 황사'로 강해졌습니다. 특히 몽골 초원의 봄풀이 아직 자라기 전이어서 흙먼지가 많았다고 중국 기상 당국은 분석했습니다. 황사로 뒤덮였던 베이징에는 다시 맑은 날씨가 찾아왔습니다. 강한 바람 때문에 모래 먼지는 하루 만에 사라졌 는데, 이번 봄에 황사가 몇 차례 더 예보가 돼있는 상황입니다. 취재기자 : 강성웅 촬영편집 : 고광 자막뉴스 : 윤현경 ▶ 기사 원문 : https://www.ytn.co.kr/_ln/0134_202103170936125411 ▶ 제보 하기 : https://mj.ytn.co.kr/mj/mj_write.php ▣ YTN 유튜브 채널 구독 : http://goo.gl/Ytb5SZ ⓒ YTN & YTN plus 무단 전재 및 재배포금지",
"video_id": "qaan1wMGnN0"
} | 16 | [
"A dust storm is forming over a desert road, with thick dust reducing visibility. The text in the image mentions a recent powerful sandstorm forming in the early morning.",
"A powerful dust storm is intensifying over a populated area, with a thick cloud of dust spreading across the sky. The text indicates that th... | qaan1wMGnN0 | A powerful dust storm is forming over a populated area in China, leading to significant consequences including the loss of 71 lives, widespread damage, and disrupted transportation. The storm poses a severe risk to human lives and has severely impacted daily life in the region. A barbed wire fence and numerous reports of civilians being trapped due to the storm are highlighted. The storm has caused extensive damage to infrastructure, including a crane that fell onto a building. A severe dust storm has reduced visibility in the area, with widespread power outages reported. |
Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China. | [
"Japan heat wave",
"\"Japan heat wave in Northeast Asia\"",
"\"Severe heat wave in North Korea\"",
"\"Northeast Asian heat wave\"",
"Record-breaking heat wave in South Korea",
"Northeast Korea heat wave",
"\"Local heat wave breaking records in China around this time\"",
"\"Extreme weather events in No... | [
"\"Japan heat wave record-breaking heat wave\"",
"\"Heat wave in Japan in Northeast Asia around [current date]\"",
"\"North Korea heat wave record-breaking heatwave during summer 2023\"",
"\"Heat wave affecting northeast asia - northeast asia\"",
"record-breaking South Korea heat wave",
"\"Northeast Asia ... | [
"\"Heatwave in Japan on a record-breaking heatwave day during the summer\"",
"Japan heatwave",
"\"Extreme weather events by name\"",
"search for \"heatwaves in northeast asia not available\"",
"\"Recent heatwave in South Korea\"",
"\"\"I'm looking for information about an upcoming event in South Korea tha... | {
"article_url": "https://ko.wikipedia.org/wiki/2018%EB%85%84_%EB%8F%99%EC%95%84%EC%8B%9C%EC%95%84_%ED%8F%AD%EC%97%BC",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nGiven the query \"Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China\", we can determine that the events that are likely to be seen in a video are related to extreme weather conditions. Considering the geographical scope of the query, which includes countries in Northeast Asia, we can infer that the video likely contains footage of heat-related phenomena.\n\nEVENTS:\n1. Heatwave\n2. Extreme weather conditions\n3. Record-breaking temperature\n4. Extreme heat\n5. Heatwave in Northeast Asia",
"en_article_excerpt": "Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.",
"en_article_url": "https://en.wikipedia.org/wiki/2018_Northeast_Asia_heat_wave",
"event_category": "disasters",
"event_name": "2018_northeast_asia_heat_wave",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nGiven the query \"Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China\", I consider the following events as reasonable candidates for extraction:\n- A severe heat wave\n- A heat wave\n- Extreme weather conditions\n- Heat wave\n- Temperature records\n- Heat wave in Asia\n\nThese events are all likely to be visible in a video related to the query and capture the essence of the query.\n\nEVENTS:\n1. Heat wave\n2. Temperature records\n3. Extreme weather conditions\n4. Weather forecast\n5. Heatwave",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nThe provided query is about a significant weather event in Northeast Asia, so it would be reasonable to extract events related to heat waves, natural disasters, and weather-related phenomena.\n\nEVENTS:\n1. Heat wave\n2. Natural disaster\n3. Weather-related event\n4. Extreme weather\n5. Heatwave",
"video_URL": "http://youtube.com/watch?v=0ZROpcl5nUE",
"video_description": "열기가 식을 새도 없이 계속 쌓여만 가고 있습니다. 폭염 특보도 확대, 강화되고 있는데요. 오늘 오전 11시를 기해 경기 남부와 강원 영서 남부에도 폭염경보가 내려졌고, 전국 대부분 지역으로 폭염주의보가 확대됐습니다. 현재 경북 경주 기온이 36.6도로 전국에서 가장 뜨겁고요. 대구 36.4도, 서울도 32.9도까지 올라 올해 최고 기온을 기록할 것으로 보입니다. 밤에도 열기는 쉽게 식지 못하고 열대야로 이어지는 곳이 많겠는데요. 내일 서울의 아침 기온 24도, 대구 25도로 오늘 밤사이에도 더위로 잠 못 이루는 곳이 많겠고요. 낮에는 다시 서울 기온이 33도, 대구 37도까지 오르는 등 오늘과 비슷하겠습니다. 현재 전국에 맑은 하늘이 드러나 있는데요. 내일도 종일 쾌청한 하늘이 함께 하겠지만, 강한 자외선에는 주의하셔야겠습니다. 당분간 뚜렷한 비 소식 없이 낮에는 폭염이, 밤에는 열대야가 기승을 부리겠습니다. 지금까지 날씨 정보였습니다. ▶ 기사 원문 : http://www.ytn.co.kr/_ln/0108_201807151449122225 ▶ 제보 안내 : http://goo.gl/gEvsAL, 모바일앱, 8585@ytn.co.kr, #2424 ▣ YTN 유튜브 채널 구독 : http://goo.gl/Ytb5SZ [ 한국 뉴스 채널 와이티엔 / Korea News Channel YTN ]",
"video_id": "0ZROpcl5nUE"
} | 16 | [
"A weather presenter is standing in front of a beach backdrop, delivering a weather report. To her left, there is a map highlighting specific areas, likely indicating weather conditions such as sunny areas.",
"The weather presenter continues her report, now with a focus on a highlighted map of South Korea, indica... | 0ZROpcl5nUE | The video is a weather report for South Korea, discussing a heatwave warning and advisory. Key actions include:
- The weather presenter delivers a weather report on a beach backdrop, highlighting areas affected by the heatwave.
- She displays maps of South Korea with temperature highlights and weather conditions, emphasizing the ongoing heatwave.
- She provides detailed forecasts for the next week, including temperature and weather conditions, using icons and numbers to indicate sunny or partly sunny skies.
- She continues to provide information and updates throughout the video, staying focused on the heatwave and its effects on the region. |
Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China. | [
"Japan heat wave",
"\"Japan heat wave in Northeast Asia\"",
"\"Severe heat wave in North Korea\"",
"\"Northeast Asian heat wave\"",
"Record-breaking heat wave in South Korea",
"Northeast Korea heat wave",
"\"Local heat wave breaking records in China around this time\"",
"\"Extreme weather events in No... | [
"\"Japan heat wave record-breaking heat wave\"",
"\"Heat wave in Japan in Northeast Asia around [current date]\"",
"\"North Korea heat wave record-breaking heatwave during summer 2023\"",
"\"Heat wave affecting northeast asia - northeast asia\"",
"record-breaking South Korea heat wave",
"\"Northeast Asia ... | [
"\"Heatwave in Japan on a record-breaking heatwave day during the summer\"",
"Japan heatwave",
"\"Extreme weather events by name\"",
"search for \"heatwaves in northeast asia not available\"",
"\"Recent heatwave in South Korea\"",
"\"\"I'm looking for information about an upcoming event in South Korea tha... | {
"article_url": "https://ko.wikipedia.org/wiki/2018%EB%85%84_%EB%8F%99%EC%95%84%EC%8B%9C%EC%95%84_%ED%8F%AD%EC%97%BC",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nGiven the query \"Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China\", we can determine that the events that are likely to be seen in a video are related to extreme weather conditions. Considering the geographical scope of the query, which includes countries in Northeast Asia, we can infer that the video likely contains footage of heat-related phenomena.\n\nEVENTS:\n1. Heatwave\n2. Extreme weather conditions\n3. Record-breaking temperature\n4. Extreme heat\n5. Heatwave in Northeast Asia",
"en_article_excerpt": "Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.",
"en_article_url": "https://en.wikipedia.org/wiki/2018_Northeast_Asia_heat_wave",
"event_category": "disasters",
"event_name": "2018_northeast_asia_heat_wave",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nGiven the query \"Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China\", I consider the following events as reasonable candidates for extraction:\n- A severe heat wave\n- A heat wave\n- Extreme weather conditions\n- Heat wave\n- Temperature records\n- Heat wave in Asia\n\nThese events are all likely to be visible in a video related to the query and capture the essence of the query.\n\nEVENTS:\n1. Heat wave\n2. Temperature records\n3. Extreme weather conditions\n4. Weather forecast\n5. Heatwave",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nThe provided query is about a significant weather event in Northeast Asia, so it would be reasonable to extract events related to heat waves, natural disasters, and weather-related phenomena.\n\nEVENTS:\n1. Heat wave\n2. Natural disaster\n3. Weather-related event\n4. Extreme weather\n5. Heatwave",
"video_URL": "http://youtube.com/watch?v=-HJV2kcDUAA",
"video_description": "서울 기온이 39도를 넘어 관측 사상 최고 기록을 갈아치운 1일 오후 서울 지하철역에서는 전력에 과부하가 걸려 스크린도어에 이상이 생겼다. 서울교통공사에 따르면 이날 오후 2시 39분께 지하철 4호선 동작역에서 당고개 방향 승강장 안전문(스크린도어)이 작동하지 않는 현상이 발생했다. 공사 측은 \"가정에서 전력에 이상이 생기면 화재 등 사고를 예방하기 위해 두꺼비집(누전차단기)이 내려가듯이, 승강장 쪽 전력에서 같은 작용이 일어난 것\"이라고 설명했다. 동작역 스크린도어는 오후 5시 30분 현재도 열려있는 상태로, 승강장에는 안전요원이 배치돼 만일의 사고에 대비하고 있다. 스크린도어가 개방된 상태일 뿐, 열차 운행에는 지장이 없는 상황이다. 공사 관계자는 \"밤에 운행을 마친 뒤에 정확한 원인을 조사할 방침\"이라면서 \"고온으로 인해 일부 부품에 문제가 생겼을 가능성도 염두에 두고 있다\"고 말했다. 영상취재 - [연합뉴스TV 차영우] ◆ 연합뉴스 홈페이지→ http://www.yna.co.kr/ ◆ 이 시각 많이 본 기사 → https://goo.gl/VQTsSZ ◆ 오늘의 핫뉴스 → https://goo.gl/WyGXpG ◆ 현장영상 → https://goo.gl/5aZcx8 ◆ 카드뉴스 →https://goo.gl/QKfDTH ◆연합뉴스 공식 SNS◆ ◇페이스북→ https://www.facebook.com/yonhap/ ▣ 연합뉴스 유튜브 채널 구독 : https://goo.gl/pL7TmT",
"video_id": "-HJV2kcDUAA"
} | 16 | [
"The image depicts a subway station platform where a train is stopped. A person wearing a black outfit and yellow socks is standing near the train doors, seemingly attempting to open or interact with the doors. Another person, wearing a high-visibility vest with an orange \"X\" pattern, stands nearby. A sign overhe... | -HJV2kcDUAA | A passenger train is stopped at a subway station due to a screen door malfunction. The screen door malfunction causes a gap between the doors and the platform, allowing people to pass through. A series of panels display information and warnings about the issue, including the possibility of high winds. The train's electrical control panel shows a red "체열림" and a yellow "장애" indicating potential failure. A train is approaching the station, and passengers are waiting, with some checking their phones or waiting for the train to stop. The train is stopped, with doors open and a gap between them, and no passengers are boarding. |
Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China. | [
"Japan heat wave",
"\"Japan heat wave in Northeast Asia\"",
"\"Severe heat wave in North Korea\"",
"\"Northeast Asian heat wave\"",
"Record-breaking heat wave in South Korea",
"Northeast Korea heat wave",
"\"Local heat wave breaking records in China around this time\"",
"\"Extreme weather events in No... | [
"\"Japan heat wave record-breaking heat wave\"",
"\"Heat wave in Japan in Northeast Asia around [current date]\"",
"\"North Korea heat wave record-breaking heatwave during summer 2023\"",
"\"Heat wave affecting northeast asia - northeast asia\"",
"record-breaking South Korea heat wave",
"\"Northeast Asia ... | [
"\"Heatwave in Japan on a record-breaking heatwave day during the summer\"",
"Japan heatwave",
"\"Extreme weather events by name\"",
"search for \"heatwaves in northeast asia not available\"",
"\"Recent heatwave in South Korea\"",
"\"\"I'm looking for information about an upcoming event in South Korea tha... | {
"article_url": "https://ko.wikipedia.org/wiki/2018%EB%85%84_%EB%8F%99%EC%95%84%EC%8B%9C%EC%95%84_%ED%8F%AD%EC%97%BC",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nGiven the query \"Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China\", we can determine that the events that are likely to be seen in a video are related to extreme weather conditions. Considering the geographical scope of the query, which includes countries in Northeast Asia, we can infer that the video likely contains footage of heat-related phenomena.\n\nEVENTS:\n1. Heatwave\n2. Extreme weather conditions\n3. Record-breaking temperature\n4. Extreme heat\n5. Heatwave in Northeast Asia",
"en_article_excerpt": "Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.",
"en_article_url": "https://en.wikipedia.org/wiki/2018_Northeast_Asia_heat_wave",
"event_category": "disasters",
"event_name": "2018_northeast_asia_heat_wave",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nGiven the query \"Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China\", I consider the following events as reasonable candidates for extraction:\n- A severe heat wave\n- A heat wave\n- Extreme weather conditions\n- Heat wave\n- Temperature records\n- Heat wave in Asia\n\nThese events are all likely to be visible in a video related to the query and capture the essence of the query.\n\nEVENTS:\n1. Heat wave\n2. Temperature records\n3. Extreme weather conditions\n4. Weather forecast\n5. Heatwave",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nThe provided query is about a significant weather event in Northeast Asia, so it would be reasonable to extract events related to heat waves, natural disasters, and weather-related phenomena.\n\nEVENTS:\n1. Heat wave\n2. Natural disaster\n3. Weather-related event\n4. Extreme weather\n5. Heatwave",
"video_URL": "http://youtube.com/watch?v=MZR5vPcBcI4",
"video_description": "극한 폭염속에도 경남 밀양 얼음골에선 주먹만한 얼음 기둥이 관측됐습니다. 신기한 건 이례적인 더위 속에서도 이 얼음이 평년보다 더 오랫동안, 꽁꽁 얼어있다는건데요. 어떻게 된 일일까요. 최진석 기자가 소개합니다. [리포트] 경남 밀양 재약산 해발 600m. 불볕더위에 바깥 온도는 36도를 넘어섰지만 바위 사이로 시원한 바람이 나옵니다. 바위 안쪽 온도를 재보니 서늘한 가을 기온인 9.2도. 바위 틈엔 주먹만한 얼음 기둥까지 있습니다. 이 지역의 명소이자 천연기념물 224호인 얼음골입니다. 불볕더위에 지친 사람들은 시원한 바람이 신기하기만 합니다. [이미경/부산광역시 북구 : \"여기까지 올라오는데 참 많이 힘들었는데, 너무 선선한 바람이 부니까 너무 너무 좋아요.\"] 바위 틈에서는 에어컨을 틀어놓은 것처럼 차가운 자연 바람이 나오고 있습니다. 열화상카메라로 온도를 재보니 1도밖에 되지 않습니다. 이곳의 얼음은 보통 장마철인 6,7월이면 다 녹아 없어지지만 올해는 계속되는 폭염의 상황에도 더 꽁꽁 얼어붙은 채 남아있습니다. [변희룡/부경대 지구환경과학과 명예교수 : \"10여 년 관측했는데, 올해 얼음이 가장 오래 관측됐거든요. 장마철에 비가 많이 오면씻겨지는데 장마철에 비가 많이 안왔어요.\"] 기록적인 폭염 속에 더위를 식히려 지난달에만 2만 여명이 찾은 밀양 얼음골. 신비로운 비밀을 간직한 채 더위에 지친 이들의 색다른 피서지가 되고 있습니다. KBS 뉴스 최진석입니다.",
"video_id": "MZR5vPcBcI4"
} | 16 | [
"The frame shows a news anchor standing in a studio, presenting a news story. The background displays an aerial view of a rocky landscape with a wooden fence, possibly indicating a recent landslide or natural disaster. The Korean text on the screen reads \"최악 폭염인데…더 '꽁꽁' 얼었다,\" which translates to \"Worst heatwave ... | MZR5vPcBcI4 | A video tells the story of a heatwave in Ilwol-gok, a region in South Korea affected by extreme temperatures. The story follows a news anchor presenting a news story, describing the unusual natural event and its contrasting effects on the region. The news anchor discusses the extreme heatwave and the area's cool temperatures, with footage of rocks, water, and people highlighting the differences between the two conditions.
The video also shows a group of people standing near a rocky area, observing the effects of the heatwave, and a group of tourists watching from behind a metal fence. A man demonstrates the difference between the extreme heat and cold by holding a flashlight and examining a crevice in rocks. The video ends with a group of people relaxing in the cold environment and a close-up view of an icy water stream. |
Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China. | [
"Japan heat wave",
"\"Japan heat wave in Northeast Asia\"",
"\"Severe heat wave in North Korea\"",
"\"Northeast Asian heat wave\"",
"Record-breaking heat wave in South Korea",
"Northeast Korea heat wave",
"\"Local heat wave breaking records in China around this time\"",
"\"Extreme weather events in No... | [
"\"Japan heat wave record-breaking heat wave\"",
"\"Heat wave in Japan in Northeast Asia around [current date]\"",
"\"North Korea heat wave record-breaking heatwave during summer 2023\"",
"\"Heat wave affecting northeast asia - northeast asia\"",
"record-breaking South Korea heat wave",
"\"Northeast Asia ... | [
"\"Heatwave in Japan on a record-breaking heatwave day during the summer\"",
"Japan heatwave",
"\"Extreme weather events by name\"",
"search for \"heatwaves in northeast asia not available\"",
"\"Recent heatwave in South Korea\"",
"\"\"I'm looking for information about an upcoming event in South Korea tha... | {
"article_url": "https://ko.wikipedia.org/wiki/2018%EB%85%84_%EB%8F%99%EC%95%84%EC%8B%9C%EC%95%84_%ED%8F%AD%EC%97%BC",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nGiven the query \"Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China\", we can determine that the events that are likely to be seen in a video are related to extreme weather conditions. Considering the geographical scope of the query, which includes countries in Northeast Asia, we can infer that the video likely contains footage of heat-related phenomena.\n\nEVENTS:\n1. Heatwave\n2. Extreme weather conditions\n3. Record-breaking temperature\n4. Extreme heat\n5. Heatwave in Northeast Asia",
"en_article_excerpt": "Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.",
"en_article_url": "https://en.wikipedia.org/wiki/2018_Northeast_Asia_heat_wave",
"event_category": "disasters",
"event_name": "2018_northeast_asia_heat_wave",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nGiven the query \"Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China\", I consider the following events as reasonable candidates for extraction:\n- A severe heat wave\n- A heat wave\n- Extreme weather conditions\n- Heat wave\n- Temperature records\n- Heat wave in Asia\n\nThese events are all likely to be visible in a video related to the query and capture the essence of the query.\n\nEVENTS:\n1. Heat wave\n2. Temperature records\n3. Extreme weather conditions\n4. Weather forecast\n5. Heatwave",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nThe provided query is about a significant weather event in Northeast Asia, so it would be reasonable to extract events related to heat waves, natural disasters, and weather-related phenomena.\n\nEVENTS:\n1. Heat wave\n2. Natural disaster\n3. Weather-related event\n4. Extreme weather\n5. Heatwave",
"video_URL": "http://youtube.com/watch?v=HCV2Hx8B0-c",
"video_description": "8월의 첫날인 오늘 최악의 폭염이 이어졌습니다. 오늘 얼마나 더웠는지 기온만 봐도 알 수가 있는데요. 강원도 홍천은 낮 기온 41도, 서울 39.6도까지 오르면서 기상 관측 이래 가장 높은 기온을 기록했습니다. 내일도 극심한 폭염은 계속됩니다. 해안가를 제외하고는 전국이 낮 기온 35도를 넘어서면서 뜨겁겠습니다. 서울과 춘천 39도, 대전과 대구는 38도까지 오르면서 오늘만큼 덥겠습니다. 토요일부터는 기온이 조금 내려가겠지만 낮 기온 35도 안팎으로 찜통더위는 이어집니다. 절기상 입추인 다음 주 화요일에도 낮 기온 35도로 무덥겠습니다. 현재 경기와 충남, 전북에는 오존주의보가 발효 중입니다. 내일도 전국이 맑겠고 자외선이 강해서 오존 농도가 높아질 수 있겠는데요. 중부와 전북 지역에서는 오존을 주의하셔야겠습니다. 아침 기온은 서울 29도로 밤 사이에도 무덥겠습니다. 한낮에는 광주 37도, 부산 35도로 오늘만큼 덥겠습니다. 바다의 물결은 제주해상에서 최고 4미터까지 거세게 일겠습니다. 더위가 밤낮을 가리지 않는 만큼 온열 질환에 걸리지 않도록 건강 관리를 잘해주시기 바랍니다. 날씨정보 전해드렸습니다.",
"video_id": "HCV2Hx8B0-c"
} | 16 | [
"The image shows a weather forecast segment featuring a female meteorologist standing in front of a backdrop that appears to be a cityscape with a clear blue sky. She is presenting weather data displayed on a graph to her left. The graph shows the projected high and low temperatures for various cities, with Seoul, ... | HCV2Hx8B0-c | A severe temperature drop is forecasted for Seoul, with decreasing temperatures of 39.6°C to 25°C over the coming days. The forecast is presented on a graph displaying a trend of decreasing temperatures in Seoul and Incheon, indicating the potential for significant drops in both minimum and maximum temperatures. A news anchor delivers news on a television broadcast, discussing economic news in the context of the forecast. |
Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China. | [
"Japan heat wave",
"\"Japan heat wave in Northeast Asia\"",
"\"Severe heat wave in North Korea\"",
"\"Northeast Asian heat wave\"",
"Record-breaking heat wave in South Korea",
"Northeast Korea heat wave",
"\"Local heat wave breaking records in China around this time\"",
"\"Extreme weather events in No... | [
"\"Japan heat wave record-breaking heat wave\"",
"\"Heat wave in Japan in Northeast Asia around [current date]\"",
"\"North Korea heat wave record-breaking heatwave during summer 2023\"",
"\"Heat wave affecting northeast asia - northeast asia\"",
"record-breaking South Korea heat wave",
"\"Northeast Asia ... | [
"\"Heatwave in Japan on a record-breaking heatwave day during the summer\"",
"Japan heatwave",
"\"Extreme weather events by name\"",
"search for \"heatwaves in northeast asia not available\"",
"\"Recent heatwave in South Korea\"",
"\"\"I'm looking for information about an upcoming event in South Korea tha... | {
"article_url": "https://ko.wikipedia.org/wiki/2018%EB%85%84_%EB%8F%99%EC%95%84%EC%8B%9C%EC%95%84_%ED%8F%AD%EC%97%BC",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nGiven the query \"Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China\", we can determine that the events that are likely to be seen in a video are related to extreme weather conditions. Considering the geographical scope of the query, which includes countries in Northeast Asia, we can infer that the video likely contains footage of heat-related phenomena.\n\nEVENTS:\n1. Heatwave\n2. Extreme weather conditions\n3. Record-breaking temperature\n4. Extreme heat\n5. Heatwave in Northeast Asia",
"en_article_excerpt": "Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.",
"en_article_url": "https://en.wikipedia.org/wiki/2018_Northeast_Asia_heat_wave",
"event_category": "disasters",
"event_name": "2018_northeast_asia_heat_wave",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nGiven the query \"Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China\", I consider the following events as reasonable candidates for extraction:\n- A severe heat wave\n- A heat wave\n- Extreme weather conditions\n- Heat wave\n- Temperature records\n- Heat wave in Asia\n\nThese events are all likely to be visible in a video related to the query and capture the essence of the query.\n\nEVENTS:\n1. Heat wave\n2. Temperature records\n3. Extreme weather conditions\n4. Weather forecast\n5. Heatwave",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nThe provided query is about a significant weather event in Northeast Asia, so it would be reasonable to extract events related to heat waves, natural disasters, and weather-related phenomena.\n\nEVENTS:\n1. Heat wave\n2. Natural disaster\n3. Weather-related event\n4. Extreme weather\n5. Heatwave",
"video_URL": "http://youtube.com/watch?v=XYrtvA02lKc",
"video_description": "경기도 분당에서 폭염으로 수도 배관이 파열되면서 교량의 균열이 발생해 교통이 통제되고 있습니다. 어젯밤(29일) 10시쯤 경기도 성남시 분당구에 있는 길이 25m, 폭 20m인 야탑 10교에 설치된 수도 배관이 터지면서 물이 치솟았습니다. 이 사고로 교각이 왼쪽으로 15도 정도 기울었고 도로 일부에도 균열이 생겼습니다. 경찰은 사고가 나자 야탑교 사거리부터 탑골 사거리까지 440m 구간 차량 통행을 전면 통제했습니다. 사고 당시 다리 위를 지나던 차량이 많지 않아 별다른 피해는 없었습니다. 경기 분당경찰서는 노후화된 교각이 폭염 영향으로 팽창하면서 짓눌린 수도관이 터져 물이 치솟은 것으로 추정된다고 설명했습니다. 최기성 [choiks7@ytn.co.kr] ▶ 기사 원문 : http://www.ytn.co.kr/_ln/0103_201807300822376551 ▶ 제보 안내 : http://goo.gl/gEvsAL, 모바일앱, 8585@ytn.co.kr, #2424 ▣ YTN 유튜브 채널 구독 : http://goo.gl/Ytb5SZ [ 한국 뉴스 채널 와이티엔 / Korea News Channel YTN ]",
"video_id": "XYrtvA02lKc"
} | 16 | [
"In the image, two news anchors, a man and a woman, are sitting at a news desk in a studio. They are dressed professionally, with the man in a dark suit and red tie, and the woman in a white top. They appear to be presenting a news segment. Behind them is a large screen displaying an image of a cityscape with a riv... | XYrtvA02lKc | The video depicts a situation where a severe heat wave is affecting a large number of apartments in Gangnam, South Korea. The video shows various scenes of apartments being affected by the heat wave, including cracked roads, damaged buildings, and closed streets. The scenes depict a road closure due to the damage caused by the heat wave, with caution tape and barriers in place to prevent vehicles from entering the affected area. The situation is described in Korean, with a news anchor discussing the severity of the heat wave and its impact on the residents and infrastructure of Gangnam.
The scenes depict the ongoing damage and the response efforts to address the situation, including the installation of warning signs and red stickers on concrete structures. The video also shows the effects of the heat wave on the surrounding environment, with greenery and trees visible in some areas, but significantly damaged in others due to the heat.
The video appears to be a news report, with the news anchors providing updates on the situation and the measures being taken to address the heat wave. The overall tone is serious and informative, highlighting the severity of the situation and the need for immediate attention to prevent further damage and ensure public safety. |
Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China. | [
"Japan heat wave",
"\"Japan heat wave in Northeast Asia\"",
"\"Severe heat wave in North Korea\"",
"\"Northeast Asian heat wave\"",
"Record-breaking heat wave in South Korea",
"Northeast Korea heat wave",
"\"Local heat wave breaking records in China around this time\"",
"\"Extreme weather events in No... | [
"\"Japan heat wave record-breaking heat wave\"",
"\"Heat wave in Japan in Northeast Asia around [current date]\"",
"\"North Korea heat wave record-breaking heatwave during summer 2023\"",
"\"Heat wave affecting northeast asia - northeast asia\"",
"record-breaking South Korea heat wave",
"\"Northeast Asia ... | [
"\"Heatwave in Japan on a record-breaking heatwave day during the summer\"",
"Japan heatwave",
"\"Extreme weather events by name\"",
"search for \"heatwaves in northeast asia not available\"",
"\"Recent heatwave in South Korea\"",
"\"\"I'm looking for information about an upcoming event in South Korea tha... | {
"article_url": "https://ko.wikipedia.org/wiki/2018%EB%85%84_%EB%8F%99%EC%95%84%EC%8B%9C%EC%95%84_%ED%8F%AD%EC%97%BC",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nGiven the query \"Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China\", we can determine that the events that are likely to be seen in a video are related to extreme weather conditions. Considering the geographical scope of the query, which includes countries in Northeast Asia, we can infer that the video likely contains footage of heat-related phenomena.\n\nEVENTS:\n1. Heatwave\n2. Extreme weather conditions\n3. Record-breaking temperature\n4. Extreme heat\n5. Heatwave in Northeast Asia",
"en_article_excerpt": "Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.",
"en_article_url": "https://en.wikipedia.org/wiki/2018_Northeast_Asia_heat_wave",
"event_category": "disasters",
"event_name": "2018_northeast_asia_heat_wave",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nGiven the query \"Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China\", I consider the following events as reasonable candidates for extraction:\n- A severe heat wave\n- A heat wave\n- Extreme weather conditions\n- Heat wave\n- Temperature records\n- Heat wave in Asia\n\nThese events are all likely to be visible in a video related to the query and capture the essence of the query.\n\nEVENTS:\n1. Heat wave\n2. Temperature records\n3. Extreme weather conditions\n4. Weather forecast\n5. Heatwave",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nThe provided query is about a significant weather event in Northeast Asia, so it would be reasonable to extract events related to heat waves, natural disasters, and weather-related phenomena.\n\nEVENTS:\n1. Heat wave\n2. Natural disaster\n3. Weather-related event\n4. Extreme weather\n5. Heatwave",
"video_URL": "http://twitter.com/i/status/1038669934375493632",
"video_description": "재난수준 폭염에도 올 여름 전력부족 없었다 올여름 재난 수준의 폭염으로 냉방 전력수요가 급증했지만 전력이 부족한 날은 없었음 7월과 8월 두 달 동안 공급예비율은 7일간을 제외, 모두 두 자릿수 이상을 유지 전력부족도 없었고, 전기세도 감면해준 문재인 정부",
"video_id": "1038669934375493632"
} | 16 | [
"The image depicts a news broadcast from a Korean news channel, where two anchors are presenting the news. The male anchor, wearing glasses, a suit, and a tie, is positioned on the left side of the screen, and the female anchor, dressed in a red blazer and black top, is on the right. The male anchor appears to be s... | 1038669934375493632 | A video presents a news segment focusing on energy management and disaster prevention in a region experiencing a severe heatwave. The video depicts a professional control room setting where engineers and technicians monitor energy systems, analyzing data and charts related to power demand, transmission lines, and energy management. The news anchors discuss and examine data on the impact of the heatwave on energy demand, highlighting the effectiveness of energy management systems in handling the situation. The video follows a continuous sequence of scenes showing the control room's setup, the engineers and technicians working, and the various data visualizations on screens. |
Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China. | [
"Japan heat wave",
"\"Japan heat wave in Northeast Asia\"",
"\"Severe heat wave in North Korea\"",
"\"Northeast Asian heat wave\"",
"Record-breaking heat wave in South Korea",
"Northeast Korea heat wave",
"\"Local heat wave breaking records in China around this time\"",
"\"Extreme weather events in No... | [
"\"Japan heat wave record-breaking heat wave\"",
"\"Heat wave in Japan in Northeast Asia around [current date]\"",
"\"North Korea heat wave record-breaking heatwave during summer 2023\"",
"\"Heat wave affecting northeast asia - northeast asia\"",
"record-breaking South Korea heat wave",
"\"Northeast Asia ... | [
"\"Heatwave in Japan on a record-breaking heatwave day during the summer\"",
"Japan heatwave",
"\"Extreme weather events by name\"",
"search for \"heatwaves in northeast asia not available\"",
"\"Recent heatwave in South Korea\"",
"\"\"I'm looking for information about an upcoming event in South Korea tha... | {
"article_url": "https://ko.wikipedia.org/wiki/2018%EB%85%84_%EB%8F%99%EC%95%84%EC%8B%9C%EC%95%84_%ED%8F%AD%EC%97%BC",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nGiven the query \"Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China\", we can determine that the events that are likely to be seen in a video are related to extreme weather conditions. Considering the geographical scope of the query, which includes countries in Northeast Asia, we can infer that the video likely contains footage of heat-related phenomena.\n\nEVENTS:\n1. Heatwave\n2. Extreme weather conditions\n3. Record-breaking temperature\n4. Extreme heat\n5. Heatwave in Northeast Asia",
"en_article_excerpt": "Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.",
"en_article_url": "https://en.wikipedia.org/wiki/2018_Northeast_Asia_heat_wave",
"event_category": "disasters",
"event_name": "2018_northeast_asia_heat_wave",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nGiven the query \"Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China\", I consider the following events as reasonable candidates for extraction:\n- A severe heat wave\n- A heat wave\n- Extreme weather conditions\n- Heat wave\n- Temperature records\n- Heat wave in Asia\n\nThese events are all likely to be visible in a video related to the query and capture the essence of the query.\n\nEVENTS:\n1. Heat wave\n2. Temperature records\n3. Extreme weather conditions\n4. Weather forecast\n5. Heatwave",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nThe provided query is about a significant weather event in Northeast Asia, so it would be reasonable to extract events related to heat waves, natural disasters, and weather-related phenomena.\n\nEVENTS:\n1. Heat wave\n2. Natural disaster\n3. Weather-related event\n4. Extreme weather\n5. Heatwave",
"video_URL": "http://twitter.com/i/status/1026325665824403456",
"video_description": "\"너무 덥지? 잘 지내고 있어?\" 관심어린 안부 한 마디가 폭염의 열기를 식힐 수 있습니다. 더위가 기승을 부리는 요즘 부모님께 또 친구에게 안부 전화 하는 건 어떨까요? (출처 : 국민안전방송 안전한TV) #성남시 #폭염 #가마솥더위 #폭염대비",
"video_id": "1026325665824403456"
} | 16 | [
"Three people are crossing a busy street, holding hands. They are dressed in professional attire, and one person is wearing a helmet while riding a motorcycle. The scene is set in a traffic-heavy urban environment, with several cars visible in the background.",
"Three people are walking along a paved path in a na... | 1026325665824403456 | Three friends walk on a paved path in a natural environment, smiling and posing together. They are dressed in professional attire. The setting is a busy urban environment. A young woman uses a microphone to communicate with another person on the phone. The conversation is ongoing, with the woman engaged in a serious discussion. The setting appears to be an indoor environment, possibly a television studio or a home office. The woman is thoughtfully engaged in the conversation, suggesting a serious discussion or interview. |
Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China. | [
"Japan heat wave",
"\"Japan heat wave in Northeast Asia\"",
"\"Severe heat wave in North Korea\"",
"\"Northeast Asian heat wave\"",
"Record-breaking heat wave in South Korea",
"Northeast Korea heat wave",
"\"Local heat wave breaking records in China around this time\"",
"\"Extreme weather events in No... | [
"\"Japan heat wave record-breaking heat wave\"",
"\"Heat wave in Japan in Northeast Asia around [current date]\"",
"\"North Korea heat wave record-breaking heatwave during summer 2023\"",
"\"Heat wave affecting northeast asia - northeast asia\"",
"record-breaking South Korea heat wave",
"\"Northeast Asia ... | [
"\"Heatwave in Japan on a record-breaking heatwave day during the summer\"",
"Japan heatwave",
"\"Extreme weather events by name\"",
"search for \"heatwaves in northeast asia not available\"",
"\"Recent heatwave in South Korea\"",
"\"\"I'm looking for information about an upcoming event in South Korea tha... | {
"article_url": "https://ko.wikipedia.org/wiki/2018%EB%85%84_%EB%8F%99%EC%95%84%EC%8B%9C%EC%95%84_%ED%8F%AD%EC%97%BC",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nGiven the query \"Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China\", we can determine that the events that are likely to be seen in a video are related to extreme weather conditions. Considering the geographical scope of the query, which includes countries in Northeast Asia, we can infer that the video likely contains footage of heat-related phenomena.\n\nEVENTS:\n1. Heatwave\n2. Extreme weather conditions\n3. Record-breaking temperature\n4. Extreme heat\n5. Heatwave in Northeast Asia",
"en_article_excerpt": "Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.",
"en_article_url": "https://en.wikipedia.org/wiki/2018_Northeast_Asia_heat_wave",
"event_category": "disasters",
"event_name": "2018_northeast_asia_heat_wave",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nGiven the query \"Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China\", I consider the following events as reasonable candidates for extraction:\n- A severe heat wave\n- A heat wave\n- Extreme weather conditions\n- Heat wave\n- Temperature records\n- Heat wave in Asia\n\nThese events are all likely to be visible in a video related to the query and capture the essence of the query.\n\nEVENTS:\n1. Heat wave\n2. Temperature records\n3. Extreme weather conditions\n4. Weather forecast\n5. Heatwave",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nThe provided query is about a significant weather event in Northeast Asia, so it would be reasonable to extract events related to heat waves, natural disasters, and weather-related phenomena.\n\nEVENTS:\n1. Heat wave\n2. Natural disaster\n3. Weather-related event\n4. Extreme weather\n5. Heatwave",
"video_URL": "http://youtube.com/watch?v=09W_DRDb9HQ",
"video_description": "+--------------------------------------------------------------------------------+ ◈ 중앙일보 공식 SNS 채널 ◈ - 페이스북 : https://www.facebook.com/joongang/ - 인스타그램 : https://www.instagram.com/joongangilbo - 트위터 : http://www.twitter.com/joongangilbo - 유튜브 : https://goo.gl/DyGoPC - 네이버 TV : http://tvcast.naver.com/joongangsisa - 네이버포스트 : https://goo.gl/eeYh4k 당신이 원하는 뉴스, 중앙일보 SNS 채널과 함께하세요. +--------------------------------------------------------------------------------+",
"video_id": "09W_DRDb9HQ"
} | 16 | [
"The image shows a piece of cheese with a part of it broken off, revealing the creamy inside. A person's hand is holding a small piece of the broken cheese. The setting appears to be a metallic surface, likely a counter or table. The text in the image provides information about a news segment from \"종합일보\" broadcas... | 09W_DRDb9HQ | A video primarily deals with the process of cooking a piece of cheese, starting with a broken piece being held by a person. It proceeds to show a series of steps, including cracking an egg into a pan and adding bacon and cheese to the pan, followed by frying the eggs and cheese. The time and context of these steps suggest the video is likely a cooking segment on a news program, with the title "중앙일보" appearing on screen. The video ends with a black screen with the title "중앙일보" again, indicating it is either continuing from a title or transition phase or possibly ending a segment. |
Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China. | [
"Japan heat wave",
"\"Japan heat wave in Northeast Asia\"",
"\"Severe heat wave in North Korea\"",
"\"Northeast Asian heat wave\"",
"Record-breaking heat wave in South Korea",
"Northeast Korea heat wave",
"\"Local heat wave breaking records in China around this time\"",
"\"Extreme weather events in No... | [
"\"Japan heat wave record-breaking heat wave\"",
"\"Heat wave in Japan in Northeast Asia around [current date]\"",
"\"North Korea heat wave record-breaking heatwave during summer 2023\"",
"\"Heat wave affecting northeast asia - northeast asia\"",
"record-breaking South Korea heat wave",
"\"Northeast Asia ... | [
"\"Heatwave in Japan on a record-breaking heatwave day during the summer\"",
"Japan heatwave",
"\"Extreme weather events by name\"",
"search for \"heatwaves in northeast asia not available\"",
"\"Recent heatwave in South Korea\"",
"\"\"I'm looking for information about an upcoming event in South Korea tha... | {
"article_url": "https://ko.wikipedia.org/wiki/2018%EB%85%84_%EB%8F%99%EC%95%84%EC%8B%9C%EC%95%84_%ED%8F%AD%EC%97%BC",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nGiven the query \"Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China\", we can determine that the events that are likely to be seen in a video are related to extreme weather conditions. Considering the geographical scope of the query, which includes countries in Northeast Asia, we can infer that the video likely contains footage of heat-related phenomena.\n\nEVENTS:\n1. Heatwave\n2. Extreme weather conditions\n3. Record-breaking temperature\n4. Extreme heat\n5. Heatwave in Northeast Asia",
"en_article_excerpt": "Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.",
"en_article_url": "https://en.wikipedia.org/wiki/2018_Northeast_Asia_heat_wave",
"event_category": "disasters",
"event_name": "2018_northeast_asia_heat_wave",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nGiven the query \"Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China\", I consider the following events as reasonable candidates for extraction:\n- A severe heat wave\n- A heat wave\n- Extreme weather conditions\n- Heat wave\n- Temperature records\n- Heat wave in Asia\n\nThese events are all likely to be visible in a video related to the query and capture the essence of the query.\n\nEVENTS:\n1. Heat wave\n2. Temperature records\n3. Extreme weather conditions\n4. Weather forecast\n5. Heatwave",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nThe provided query is about a significant weather event in Northeast Asia, so it would be reasonable to extract events related to heat waves, natural disasters, and weather-related phenomena.\n\nEVENTS:\n1. Heat wave\n2. Natural disaster\n3. Weather-related event\n4. Extreme weather\n5. Heatwave",
"video_URL": "http://youtube.com/watch?v=8wcKX0LCc08",
"video_description": "폭염 때문에 힘든 건 사람만이 아닙니다. 과육이 물러진 수박이 산지에서 폐기되는 등 제철 과일과 채소 피해도 잇따르고 있습니다.",
"video_id": "8wcKX0LCc08"
} | 16 | [
"A news anchor is standing in front of a screen displaying a field of watermelons, some of which are damaged or rotting. The text on the screen reads \"수박발 갈아엎어...제철 과일 직격탄\" which translates to \"Watermelon fields destroyed...Summer fruit hit hard.\" The image also shows a timestamp of 07:18 and the channel name \... | 8wcKX0LCc08 | A severe heatwave has affected watermelon and corn fields in various parts of South Korea, causing damage and power outages. The power outage has resulted in significant yield reductions in these crops. Key figures, including news anchors and farmers, have been interviewed, providing insight into the situation. Agricultural products are affected, with many households being impacted by the power outage. |
Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China. | [
"Japan heat wave",
"\"Japan heat wave in Northeast Asia\"",
"\"Severe heat wave in North Korea\"",
"\"Northeast Asian heat wave\"",
"Record-breaking heat wave in South Korea",
"Northeast Korea heat wave",
"\"Local heat wave breaking records in China around this time\"",
"\"Extreme weather events in No... | [
"\"Japan heat wave record-breaking heat wave\"",
"\"Heat wave in Japan in Northeast Asia around [current date]\"",
"\"North Korea heat wave record-breaking heatwave during summer 2023\"",
"\"Heat wave affecting northeast asia - northeast asia\"",
"record-breaking South Korea heat wave",
"\"Northeast Asia ... | [
"\"Heatwave in Japan on a record-breaking heatwave day during the summer\"",
"Japan heatwave",
"\"Extreme weather events by name\"",
"search for \"heatwaves in northeast asia not available\"",
"\"Recent heatwave in South Korea\"",
"\"\"I'm looking for information about an upcoming event in South Korea tha... | {
"article_url": "https://ko.wikipedia.org/wiki/2018%EB%85%84_%EB%8F%99%EC%95%84%EC%8B%9C%EC%95%84_%ED%8F%AD%EC%97%BC",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nGiven the query \"Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China\", we can determine that the events that are likely to be seen in a video are related to extreme weather conditions. Considering the geographical scope of the query, which includes countries in Northeast Asia, we can infer that the video likely contains footage of heat-related phenomena.\n\nEVENTS:\n1. Heatwave\n2. Extreme weather conditions\n3. Record-breaking temperature\n4. Extreme heat\n5. Heatwave in Northeast Asia",
"en_article_excerpt": "Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.",
"en_article_url": "https://en.wikipedia.org/wiki/2018_Northeast_Asia_heat_wave",
"event_category": "disasters",
"event_name": "2018_northeast_asia_heat_wave",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\nGiven the query \"Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China\", I consider the following events as reasonable candidates for extraction:\n- A severe heat wave\n- A heat wave\n- Extreme weather conditions\n- Heat wave\n- Temperature records\n- Heat wave in Asia\n\nThese events are all likely to be visible in a video related to the query and capture the essence of the query.\n\nEVENTS:\n1. Heat wave\n2. Temperature records\n3. Extreme weather conditions\n4. Weather forecast\n5. Heatwave",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Throughout much of July 2018, a record-breaking heat wave affected large areas of Northeast Asia including Japan, North Korea, South Korea, China.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nThe provided query is about a significant weather event in Northeast Asia, so it would be reasonable to extract events related to heat waves, natural disasters, and weather-related phenomena.\n\nEVENTS:\n1. Heat wave\n2. Natural disaster\n3. Weather-related event\n4. Extreme weather\n5. Heatwave",
"video_URL": "http://youtube.com/watch?v=g9CbgCp52qc",
"video_description": "연일 불볕더위가 기승을 부리는 가운데 서울시가 도로에 물을 뿌려 열을 식히는 작업을 강화합니다. 서울시는 낮 최고기온이 32℃ 이상으로 오르면 폭염에 노출될 가능성이 큰 정오부터 오후 4시까지 중앙버스전용차로를 중심으로 살수 차량 운행을 확대한다고 밝혔습니다. 서울시는 여름철 도로 물뿌리기로 미세먼지를 제거하는 동시에 온열질환 등 폭염으로 발생할 수 있는 안전사고도 방지할 수 있기를 기대하고 있습니다. ▶ 기사 원문 : http://www.ytn.co.kr/_ln/0115_201807181131189139 ▶ 제보 안내 : http://goo.gl/gEvsAL, 모바일앱, 8585@ytn.co.kr, #2424 ▣ YTN 유튜브 채널 구독 : http://goo.gl/Ytb5SZ [ 한국 뉴스 채널 와이티엔 / Korea News Channel YTN ]",
"video_id": "g9CbgCp52qc"
} | 16 | [
"In the current frame, two news anchors are seated at a desk in a studio. The male anchor on the left is wearing a suit and tie and appears to be speaking or preparing to speak, with papers in front of him. The female anchor on the right is dressed in a pink blouse and is also facing the camera. Behind them is a la... | g9CbgCp52qc | A video about the ongoing effort to combat heatwaves by watering streets, focusing on urban environments. The scenes show a large truck spraying water on a busy city street, and smaller vehicles and pedestrians reacting to the action. The video aims to emphasize the measures taken to manage the heat in cities by watering roads, with the truck driving through a city during this process. |
In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions. | [
"\"Snowfall in East Asia\"",
"Weather advisory for parts of Southeast Asia and northern South Asia",
"\"Snowfall in East Asia during a cold wave\"",
"\"Temperature trends in East Asia, Southeast Asia, and South Asia during the winter season\"",
"\"Cold temperatures in Parts of Southeast Asia and Northern So... | [] | [
"\"cold wave snowfall in east asia\"",
"\"When parts of Southeast Asia experience cold wave in winter/spring/autumn, I'd like information about snowfall in specific sub-regions.\"",
"\"Cold wave hits Southeast Asia, happening in late winter, with heavy snowfall in major cities.\"",
"\"Cold weather records in ... | {
"article_url": "https://ko.wikipedia.org/wiki/2016%EB%85%84_1%EC%9B%94_%EB%8F%99%EC%95%84%EC%8B%9C%EC%95%84_%ED%95%9C%ED%8C%8C",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nGiven the query \"In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions,\" the events that make sense to extract are those related to extreme weather conditions, specifically cold temperatures and snowfall. This implies that the query is likely to be visualized in a video that depicts these weather events.\n\nEVENTS:\n1. Snowfall in Southeast Asia\n2. Record cold temperatures in East Asia\n3. Protests in Hong Kong (although this might be considered subjective, as protests in Hong Kong in January 2016 are possible but less likely compared to more recent events)\n4. Widespread cold wave across Asia\n5. Record-breaking cold temperatures",
"en_article_excerpt": "In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions.",
"en_article_url": "https://en.wikipedia.org/wiki/January_2016_East_Asia_cold_wave",
"event_category": "disasters",
"event_name": "2016_east_asia_cold_wave",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "\"EXPLANATION:\"\nThe events that make sense to extract for the query are related to weather and natural disasters. Given the context of a cold wave, it makes sense to extract events that describe the cold temperatures and snowfall. This can help create a visual representation of the event in a video, possibly showing scenes of people covering themselves with layers, looking at the sky, or searching for shelter.\n\nEVENTS:\n1. Snowfall\n2. Cold temperatures\n3. Record cold wave\n4. Part of the cold wave\n5. Record temperatures",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nBased on my analysis of the given query, I determined that it is highly likely to be an event that could be visualized in a video. The query mentions a cold wave, which is a clear indicator of a weather-related event. The mention of snowfall, a specific and concrete weather phenomenon, also supports the idea of an event that could be visualized in a video. Therefore, it is reasonable to expect that the events extracted from the query would be weather-related. \n\nEvents:\n1. Snowfall\n2. Temperature drop\n3. Cold wave",
"video_URL": "http://twitter.com/i/status/1346958705632296961",
"video_description": "전국 눈 대단한 한파 다들 잘 이겨내요",
"video_id": "1346958705632296961"
} | 16 | [
"The image depicts a residential area with high-rise apartment buildings on either side of a central pathway or park area. The buildings are modern, multi-story structures with a uniform design. The central area appears to be a landscaped park with trees and pathways, though visibility is limited due to fog or low-... | 1346958705632296961 | A group of residents live in a series of high-rise apartment buildings and a central park in a residential area. The park is landscaped with trees and pathways, but visibility is limited due to fog. The residents walk along the park's pathways, and vehicles occasionally pass by, but mostly remain still. The scene captures a tranquil atmosphere, with only minor elements of activity, such as a red vehicle on the road near the buildings. |
In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions. | [
"\"Snowfall in East Asia\"",
"Weather advisory for parts of Southeast Asia and northern South Asia",
"\"Snowfall in East Asia during a cold wave\"",
"\"Temperature trends in East Asia, Southeast Asia, and South Asia during the winter season\"",
"\"Cold temperatures in Parts of Southeast Asia and Northern So... | [] | [
"\"cold wave snowfall in east asia\"",
"\"When parts of Southeast Asia experience cold wave in winter/spring/autumn, I'd like information about snowfall in specific sub-regions.\"",
"\"Cold wave hits Southeast Asia, happening in late winter, with heavy snowfall in major cities.\"",
"\"Cold weather records in ... | {
"article_url": "https://ko.wikipedia.org/wiki/2016%EB%85%84_1%EC%9B%94_%EB%8F%99%EC%95%84%EC%8B%9C%EC%95%84_%ED%95%9C%ED%8C%8C",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nGiven the query \"In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions,\" the events that make sense to extract are those related to extreme weather conditions, specifically cold temperatures and snowfall. This implies that the query is likely to be visualized in a video that depicts these weather events.\n\nEVENTS:\n1. Snowfall in Southeast Asia\n2. Record cold temperatures in East Asia\n3. Protests in Hong Kong (although this might be considered subjective, as protests in Hong Kong in January 2016 are possible but less likely compared to more recent events)\n4. Widespread cold wave across Asia\n5. Record-breaking cold temperatures",
"en_article_excerpt": "In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions.",
"en_article_url": "https://en.wikipedia.org/wiki/January_2016_East_Asia_cold_wave",
"event_category": "disasters",
"event_name": "2016_east_asia_cold_wave",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "\"EXPLANATION:\"\nThe events that make sense to extract for the query are related to weather and natural disasters. Given the context of a cold wave, it makes sense to extract events that describe the cold temperatures and snowfall. This can help create a visual representation of the event in a video, possibly showing scenes of people covering themselves with layers, looking at the sky, or searching for shelter.\n\nEVENTS:\n1. Snowfall\n2. Cold temperatures\n3. Record cold wave\n4. Part of the cold wave\n5. Record temperatures",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nBased on my analysis of the given query, I determined that it is highly likely to be an event that could be visualized in a video. The query mentions a cold wave, which is a clear indicator of a weather-related event. The mention of snowfall, a specific and concrete weather phenomenon, also supports the idea of an event that could be visualized in a video. Therefore, it is reasonable to expect that the events extracted from the query would be weather-related. \n\nEvents:\n1. Snowfall\n2. Temperature drop\n3. Cold wave",
"video_URL": "http://youtube.com/watch?v=2qCY0mBeZy4",
"video_description": "지난 연말부터 시작된 한파의 기세는 여전했습니다. 강원도 산간은 영하 20도, 철원 영하 15.9도, 서울도 영하 9.8도를 기록했습니다. 이번 한파는 풀리지 않고 다음 주까지 이어질 것으로 보입니다. 주말부터 다음 주 초반까지 기온이 조금 오르지만, 이후 다시 기온이 급강하하기 때문입니다. 특히 다음 주 후반에는 서울 기온이 영하 15도까지 떨어지는 강력한 한파가 맹위를 떨칠 전망입니다. 올겨울 들어 가장 강력한 추위입니다. [윤기한 / 기상청 통보관 : 한파가 누그러지지 않은 상황에서 2차로 한기가 추가로 내려오는 것입니다. 따라서 다음 주에는 추위 강도가 지금보다 훨씬 세질 것으로 보입니다.] 북극 한파가 새해 첫날부터 열흘 이상 이어지는 것입니다. 서해안의 눈도 잠시 약해졌다가 다음 주 중반 이후에는 눈 폭탄으로 쏟아질 것으로 우려됩니다. 기상청은 이번 한파와 폭설로 동파와 축사, 비닐하우스 붕괴 등 시설물 피해가 속출할 것이라며 사전 대비를 당부했습니다. 취재기자: 김진두 영상편집: 한경희 자막뉴스: 박해진 ▶ 기사 원문 : https://www.ytn.co.kr/replay/view.php?idx=29&key=202101020712296278 ▶ 제보 안내 : http://goo.gl/gEvsAL, 모바일앱, social@ytn.co.kr, #2424 ▣ YTN 유튜브 채널 구독 : http://goo.gl/Ytb5SZ ⓒ YTN & YTN plus 무단 전재 및 재배포금지",
"video_id": "2qCY0mBeZy4"
} | 16 | [
"The image depicts a sunrise or sunset with sunlight streaming through a hazy, misty environment. The scene includes tall, dry grass in the foreground and a backdrop of rolling hills partially shrouded in mist. The Korean text in the bottom left corner reads, \"지난 연말부터 시작된 한파의 기세는 여전합니다.\" which translates to \"The... | 2qCY0mBeZy4 | A severe cold wave, referred to as the "지난 연말부터 시작된 한파의 기세" or "last year's cold wave," has been ongoing for several weeks. The cold wave has been marked by severe cold temperatures, with regions experiencing temperatures as low as -20°C in Gangwon-do and -15.9°C in Cheorwon. The cold wave persisted for an extended period, with temperatures fluctuating and rising slightly in some areas until the weekend, only to drop sharply again. The cold wave has had a significant impact on daily life, with people bundled up in winter clothing, including fur-lined coats and masks, and snow-covered roads becoming hazardous to travel. The cold wave has caused widespread damage and disruption to daily activities, including the collapse of facilities such as buses and houses. The Korea Meteorological Administration has issued warnings and safety advisories regarding the severe weather conditions, emphasizing the need for preparedness and caution. The cold wave is expected to continue for an extended period, with temperatures expected to rise significantly in the coming days. |
In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions. | [
"\"Snowfall in East Asia\"",
"Weather advisory for parts of Southeast Asia and northern South Asia",
"\"Snowfall in East Asia during a cold wave\"",
"\"Temperature trends in East Asia, Southeast Asia, and South Asia during the winter season\"",
"\"Cold temperatures in Parts of Southeast Asia and Northern So... | [] | [
"\"cold wave snowfall in east asia\"",
"\"When parts of Southeast Asia experience cold wave in winter/spring/autumn, I'd like information about snowfall in specific sub-regions.\"",
"\"Cold wave hits Southeast Asia, happening in late winter, with heavy snowfall in major cities.\"",
"\"Cold weather records in ... | {
"article_url": "https://ko.wikipedia.org/wiki/2016%EB%85%84_1%EC%9B%94_%EB%8F%99%EC%95%84%EC%8B%9C%EC%95%84_%ED%95%9C%ED%8C%8C",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nGiven the query \"In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions,\" the events that make sense to extract are those related to extreme weather conditions, specifically cold temperatures and snowfall. This implies that the query is likely to be visualized in a video that depicts these weather events.\n\nEVENTS:\n1. Snowfall in Southeast Asia\n2. Record cold temperatures in East Asia\n3. Protests in Hong Kong (although this might be considered subjective, as protests in Hong Kong in January 2016 are possible but less likely compared to more recent events)\n4. Widespread cold wave across Asia\n5. Record-breaking cold temperatures",
"en_article_excerpt": "In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions.",
"en_article_url": "https://en.wikipedia.org/wiki/January_2016_East_Asia_cold_wave",
"event_category": "disasters",
"event_name": "2016_east_asia_cold_wave",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "\"EXPLANATION:\"\nThe events that make sense to extract for the query are related to weather and natural disasters. Given the context of a cold wave, it makes sense to extract events that describe the cold temperatures and snowfall. This can help create a visual representation of the event in a video, possibly showing scenes of people covering themselves with layers, looking at the sky, or searching for shelter.\n\nEVENTS:\n1. Snowfall\n2. Cold temperatures\n3. Record cold wave\n4. Part of the cold wave\n5. Record temperatures",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nBased on my analysis of the given query, I determined that it is highly likely to be an event that could be visualized in a video. The query mentions a cold wave, which is a clear indicator of a weather-related event. The mention of snowfall, a specific and concrete weather phenomenon, also supports the idea of an event that could be visualized in a video. Therefore, it is reasonable to expect that the events extracted from the query would be weather-related. \n\nEvents:\n1. Snowfall\n2. Temperature drop\n3. Cold wave",
"video_URL": "http://youtube.com/watch?v=QmU3W9ffZeM",
"video_description": "기온이 마치 롤러코스터처럼 떨어질 모양입니다. 얼마나 추워질지 벌써부터 겁도 나는데요. 일단 내일 아침 서울 기온은 영하 11도까지 떨어지겠고요. 목요일부터는 찬바람이 더욱 강해지면서 아침 기온이 영하 15도로 전날 낮 대비 13도가량 기온이 급감하겠습니다. 한낮 기온도 영하 11도에 머물 텐데요. 다음 날 금요일 아침 기온은 영하 17도로 이번 한파의 절정이 되겠습니다. 한파 대비 미리 해주시는 게 좋겠습니다. 이에 따라서 오늘 밤 9시를 기해서는 경기 동부와 북부, 충청도, 경북 곳곳에 한파경보가 발효되겠고요. 서울 등 그 밖의 수도권과 충청도 전북 내륙 지역까지 한파주의보가 확대 발효되겠습니다. 눈구름도 함께 몰려오겠습니다. 내일 퇴근길 무렵에 서해 등 서쪽지역에 눈이 내리기 시작해 밤사이에 전국으로 확대될 텐데요. 특히 서해안 지역으로는 최대 30cm 이상, 제주 산지에는 최고 50cm 이상의 폭설이 우려됩니다. 충남 지역에는 금요일까지 호남과 제주도는 일요일까지 눈이 길게 이어지겠습니다. 날씨였습니다. https://imnews.imbc.com/replay/2021/nw1700/article/6048848_34922.html #날씨,#추위 #한파",
"video_id": "QmU3W9ffZeM"
} | 16 | [
"A woman in a brown dress stands by a river with grass in the background, gesturing as if she is reporting or presenting.",
"The woman continues her report in the same location by the river, pointing towards a digital weather chart that displays temperature changes over four days, indicating a drop from -3.6°C to... | QmU3W9ffZeM | A woman reports the weather by the river, describing a series of temperature drops from -3.6°C to -17°C over four days, accompanied by a digital weather chart displaying the forecast. She highlights areas affected by the cold, including regions in blue, and explains the impact on those areas. The forecast indicates heavy snowfall starting tonight, with an expected accumulation of at least 30 cm, emphasizing the ongoing weather pattern. |
In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions. | [
"\"Snowfall in East Asia\"",
"Weather advisory for parts of Southeast Asia and northern South Asia",
"\"Snowfall in East Asia during a cold wave\"",
"\"Temperature trends in East Asia, Southeast Asia, and South Asia during the winter season\"",
"\"Cold temperatures in Parts of Southeast Asia and Northern So... | [] | [
"\"cold wave snowfall in east asia\"",
"\"When parts of Southeast Asia experience cold wave in winter/spring/autumn, I'd like information about snowfall in specific sub-regions.\"",
"\"Cold wave hits Southeast Asia, happening in late winter, with heavy snowfall in major cities.\"",
"\"Cold weather records in ... | {
"article_url": "https://ko.wikipedia.org/wiki/2016%EB%85%84_1%EC%9B%94_%EB%8F%99%EC%95%84%EC%8B%9C%EC%95%84_%ED%95%9C%ED%8C%8C",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nGiven the query \"In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions,\" the events that make sense to extract are those related to extreme weather conditions, specifically cold temperatures and snowfall. This implies that the query is likely to be visualized in a video that depicts these weather events.\n\nEVENTS:\n1. Snowfall in Southeast Asia\n2. Record cold temperatures in East Asia\n3. Protests in Hong Kong (although this might be considered subjective, as protests in Hong Kong in January 2016 are possible but less likely compared to more recent events)\n4. Widespread cold wave across Asia\n5. Record-breaking cold temperatures",
"en_article_excerpt": "In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions.",
"en_article_url": "https://en.wikipedia.org/wiki/January_2016_East_Asia_cold_wave",
"event_category": "disasters",
"event_name": "2016_east_asia_cold_wave",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "\"EXPLANATION:\"\nThe events that make sense to extract for the query are related to weather and natural disasters. Given the context of a cold wave, it makes sense to extract events that describe the cold temperatures and snowfall. This can help create a visual representation of the event in a video, possibly showing scenes of people covering themselves with layers, looking at the sky, or searching for shelter.\n\nEVENTS:\n1. Snowfall\n2. Cold temperatures\n3. Record cold wave\n4. Part of the cold wave\n5. Record temperatures",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nBased on my analysis of the given query, I determined that it is highly likely to be an event that could be visualized in a video. The query mentions a cold wave, which is a clear indicator of a weather-related event. The mention of snowfall, a specific and concrete weather phenomenon, also supports the idea of an event that could be visualized in a video. Therefore, it is reasonable to expect that the events extracted from the query would be weather-related. \n\nEvents:\n1. Snowfall\n2. Temperature drop\n3. Cold wave",
"video_URL": "http://youtube.com/watch?v=G7XGz4uL6mI",
"video_description": "기상정보 배혜지 기상캐스터 2021년 1월 12일 화요일 아침 시각 빙판길 골목길 언덕길 등 보행 조심하세요... 감기 걸리지 않도록 건강관리 잘하세요. 행복한 화요일 되세요.^^ #감기조심 #한파특보 #한파경보 #한파주의보 #건조경보 #화재예방 #산불조심 #코로나19 #COVID19 #기상정보 #날씨 #Weather #배혜지기상캐스터 #기상캐스터 #WeatherCaster #KBS뉴스광장 ============================================================== Uploaded by: 하안마음 (White Heart.rhythm7777777) Youtube: http://www.youtube.com/user/rhythmsgarden Twitter: https://twitter.com/rhythm700 Face Book: https://www.facebook.com/rhythm700 Instagram: https://www.instagram.com/rhythm7777777 헤일리 웨스튼라 Hayley Westenra Official Website: http://www.hayleywestenra.com Instagram: https://www.instagram.com/hayleywestenra Twitter: https://twitter.com/hayleywestenra FaceBook: https://www.facebook.com/hayleywestenra Youtube: https://www.youtube.com/channel/UCuA623UAgxrJuWcMcaI_kgg 헤일리 웨스튼라 코리아 Hayley Westenra Korea Fancafe: http://cafe.daum.net/HayleyWestenra Facebook: https://www.facebook.com/hayleywestenra Youtube: https://www.youtube.com/channel/UCBrwEte0dnl1JTWSpzGFVeA/videos Wikipedia: http://en.wikipedia.org/wiki/Hayley_Westenra",
"video_id": "G7XGz4uL6mI"
} | 16 | [
"The frame features a weather forecast segment on a news channel, KBS. The presenter, dressed in a yellow dress, stands on a snowy background with snow-covered mountains in the backdrop. She is providing information about the temperature, specifically noting a significant increase from the previous day, as indicate... | G7XGz4uL6mI | The video depicts a weather forecast segment on a Korean news channel, KBS. The presenter discusses a significant increase in temperature in a mountainous region, with the current temperature being 1.0°C higher than the previous day's -9.6°C. The presenter explains the increase in temperature, using visual aids to inform viewers about the recent changes in temperature. The forecast segment is set in a snowy mountainous region and later transitions to a hazy cityscape, where a fine dust forecast is shown, indicating areas with high levels of fine dust. The presenter explains the current fine dust levels and the affected regions, emphasizing the importance of air quality. The forecast segment continues to cover various weather conditions, including a radar video showing potential weather activity, snowfall regions, and time intervals where snow is expected. The broadcast time remains constant throughout, indicating that the segment is continuous and informative. |
In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions. | [
"\"Snowfall in East Asia\"",
"Weather advisory for parts of Southeast Asia and northern South Asia",
"\"Snowfall in East Asia during a cold wave\"",
"\"Temperature trends in East Asia, Southeast Asia, and South Asia during the winter season\"",
"\"Cold temperatures in Parts of Southeast Asia and Northern So... | [] | [
"\"cold wave snowfall in east asia\"",
"\"When parts of Southeast Asia experience cold wave in winter/spring/autumn, I'd like information about snowfall in specific sub-regions.\"",
"\"Cold wave hits Southeast Asia, happening in late winter, with heavy snowfall in major cities.\"",
"\"Cold weather records in ... | {
"article_url": "https://ko.wikipedia.org/wiki/2016%EB%85%84_1%EC%9B%94_%EB%8F%99%EC%95%84%EC%8B%9C%EC%95%84_%ED%95%9C%ED%8C%8C",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nGiven the query \"In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions,\" the events that make sense to extract are those related to extreme weather conditions, specifically cold temperatures and snowfall. This implies that the query is likely to be visualized in a video that depicts these weather events.\n\nEVENTS:\n1. Snowfall in Southeast Asia\n2. Record cold temperatures in East Asia\n3. Protests in Hong Kong (although this might be considered subjective, as protests in Hong Kong in January 2016 are possible but less likely compared to more recent events)\n4. Widespread cold wave across Asia\n5. Record-breaking cold temperatures",
"en_article_excerpt": "In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions.",
"en_article_url": "https://en.wikipedia.org/wiki/January_2016_East_Asia_cold_wave",
"event_category": "disasters",
"event_name": "2016_east_asia_cold_wave",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "\"EXPLANATION:\"\nThe events that make sense to extract for the query are related to weather and natural disasters. Given the context of a cold wave, it makes sense to extract events that describe the cold temperatures and snowfall. This can help create a visual representation of the event in a video, possibly showing scenes of people covering themselves with layers, looking at the sky, or searching for shelter.\n\nEVENTS:\n1. Snowfall\n2. Cold temperatures\n3. Record cold wave\n4. Part of the cold wave\n5. Record temperatures",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nBased on my analysis of the given query, I determined that it is highly likely to be an event that could be visualized in a video. The query mentions a cold wave, which is a clear indicator of a weather-related event. The mention of snowfall, a specific and concrete weather phenomenon, also supports the idea of an event that could be visualized in a video. Therefore, it is reasonable to expect that the events extracted from the query would be weather-related. \n\nEvents:\n1. Snowfall\n2. Temperature drop\n3. Cold wave",
"video_URL": "http://twitter.com/i/status/1346762056142188546",
"video_description": "눈보라 때문에 기절을 함",
"video_id": "1346762056142188546"
} | 16 | [
"Heavy snowfall is occurring on a city street, blanketing the ground and buildings. A building with the sign \"JJD HAIR\" and a neon red light along the roofline is visible. The street appears mostly empty due to the weather conditions.",
"The heavy snowfall continues, covering the street and surrounding building... | 1346762056142188546 | A severe snowstorm is depicted in a single location, with a building called "JJD HAIR" and neon red light on the roofline. The weather conditions are severe, with heavy snowfall blanketing the street and surrounding buildings. The presence of a deserted street, with no pedestrians or vehicles, highlights the severity of the weather. |
In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions. | [
"\"Snowfall in East Asia\"",
"Weather advisory for parts of Southeast Asia and northern South Asia",
"\"Snowfall in East Asia during a cold wave\"",
"\"Temperature trends in East Asia, Southeast Asia, and South Asia during the winter season\"",
"\"Cold temperatures in Parts of Southeast Asia and Northern So... | [] | [
"\"cold wave snowfall in east asia\"",
"\"When parts of Southeast Asia experience cold wave in winter/spring/autumn, I'd like information about snowfall in specific sub-regions.\"",
"\"Cold wave hits Southeast Asia, happening in late winter, with heavy snowfall in major cities.\"",
"\"Cold weather records in ... | {
"article_url": "https://ko.wikipedia.org/wiki/2016%EB%85%84_1%EC%9B%94_%EB%8F%99%EC%95%84%EC%8B%9C%EC%95%84_%ED%95%9C%ED%8C%8C",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nGiven the query \"In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions,\" the events that make sense to extract are those related to extreme weather conditions, specifically cold temperatures and snowfall. This implies that the query is likely to be visualized in a video that depicts these weather events.\n\nEVENTS:\n1. Snowfall in Southeast Asia\n2. Record cold temperatures in East Asia\n3. Protests in Hong Kong (although this might be considered subjective, as protests in Hong Kong in January 2016 are possible but less likely compared to more recent events)\n4. Widespread cold wave across Asia\n5. Record-breaking cold temperatures",
"en_article_excerpt": "In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions.",
"en_article_url": "https://en.wikipedia.org/wiki/January_2016_East_Asia_cold_wave",
"event_category": "disasters",
"event_name": "2016_east_asia_cold_wave",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "\"EXPLANATION:\"\nThe events that make sense to extract for the query are related to weather and natural disasters. Given the context of a cold wave, it makes sense to extract events that describe the cold temperatures and snowfall. This can help create a visual representation of the event in a video, possibly showing scenes of people covering themselves with layers, looking at the sky, or searching for shelter.\n\nEVENTS:\n1. Snowfall\n2. Cold temperatures\n3. Record cold wave\n4. Part of the cold wave\n5. Record temperatures",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nBased on my analysis of the given query, I determined that it is highly likely to be an event that could be visualized in a video. The query mentions a cold wave, which is a clear indicator of a weather-related event. The mention of snowfall, a specific and concrete weather phenomenon, also supports the idea of an event that could be visualized in a video. Therefore, it is reasonable to expect that the events extracted from the query would be weather-related. \n\nEvents:\n1. Snowfall\n2. Temperature drop\n3. Cold wave",
"video_URL": "http://twitter.com/i/status/1346961474866925570",
"video_description": "눈보라가 휘날리는 바람 찬 부산 시청 #풍산",
"video_id": "1346961474866925570"
} | 16 | [
"A person wearing a black coat and gloves is standing on a sidewalk near a street intersection, holding a green board or sign. The area appears to be a city street with several orange traffic cones placed on the road, possibly indicating some form of roadwork or traffic management. In the background, there are buil... | 1346961474866925570 | A group of protesters gather near a city intersection, holding signs and engaging in a peaceful demonstration or vigil. The event appears to be related to an organized protest or demonstration, likely related to a social or environmental issue, given the presence of leafless trees and snow-covered sidewalks. The protesters are arranged in a loose formation, with some standing near a pedestrian crossing and others on the sidewalk, holding their signs. The scene is set in a cold, wintry environment with traffic cones and snow-covered traffic. The protesters' actions appear to be peaceful, with no apparent confrontation or aggressive behavior. |
In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions. | [
"\"Snowfall in East Asia\"",
"Weather advisory for parts of Southeast Asia and northern South Asia",
"\"Snowfall in East Asia during a cold wave\"",
"\"Temperature trends in East Asia, Southeast Asia, and South Asia during the winter season\"",
"\"Cold temperatures in Parts of Southeast Asia and Northern So... | [] | [
"\"cold wave snowfall in east asia\"",
"\"When parts of Southeast Asia experience cold wave in winter/spring/autumn, I'd like information about snowfall in specific sub-regions.\"",
"\"Cold wave hits Southeast Asia, happening in late winter, with heavy snowfall in major cities.\"",
"\"Cold weather records in ... | {
"article_url": "https://ko.wikipedia.org/wiki/2016%EB%85%84_1%EC%9B%94_%EB%8F%99%EC%95%84%EC%8B%9C%EC%95%84_%ED%95%9C%ED%8C%8C",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\nGiven the query \"In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions,\" the events that make sense to extract are those related to extreme weather conditions, specifically cold temperatures and snowfall. This implies that the query is likely to be visualized in a video that depicts these weather events.\n\nEVENTS:\n1. Snowfall in Southeast Asia\n2. Record cold temperatures in East Asia\n3. Protests in Hong Kong (although this might be considered subjective, as protests in Hong Kong in January 2016 are possible but less likely compared to more recent events)\n4. Widespread cold wave across Asia\n5. Record-breaking cold temperatures",
"en_article_excerpt": "In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions.",
"en_article_url": "https://en.wikipedia.org/wiki/January_2016_East_Asia_cold_wave",
"event_category": "disasters",
"event_name": "2016_east_asia_cold_wave",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "\"EXPLANATION:\"\nThe events that make sense to extract for the query are related to weather and natural disasters. Given the context of a cold wave, it makes sense to extract events that describe the cold temperatures and snowfall. This can help create a visual representation of the event in a video, possibly showing scenes of people covering themselves with layers, looking at the sky, or searching for shelter.\n\nEVENTS:\n1. Snowfall\n2. Cold temperatures\n3. Record cold wave\n4. Part of the cold wave\n5. Record temperatures",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: In late January 2016, a cold wave struck much of East Asia, parts of mainland Southeast Asia and parts of northern South Asia, bringing record cold temperatures and snowfall to many regions.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nBased on my analysis of the given query, I determined that it is highly likely to be an event that could be visualized in a video. The query mentions a cold wave, which is a clear indicator of a weather-related event. The mention of snowfall, a specific and concrete weather phenomenon, also supports the idea of an event that could be visualized in a video. Therefore, it is reasonable to expect that the events extracted from the query would be weather-related. \n\nEvents:\n1. Snowfall\n2. Temperature drop\n3. Cold wave",
"video_URL": "http://youtube.com/watch?v=PxIbNGPQYw8",
"video_description": "지난주 부터 계속되는 한파에 SNS 등에서는 순간포착 영상인 이른바 '얼었어' 시리즈가 유행하고 있다고 합니다. https://imnews.imbc.com/replay/2021/nwtoday/article/6054105_34943.html #한파, #SNS, #얼었어시리즈",
"video_id": "PxIbNGPQYw8"
} | 16 | [
"The image depicts a news broadcast with two male anchors standing in a studio. The background displays Korean text on a screen, indicating a news segment. The anchors are dressed formally, with one holding papers. The text on the screen includes headlines and information about current events, such as a discussion ... | PxIbNGPQYw8 | A video discusses the impact of severe cold weather on various aspects of daily life, including economics, consumer behavior, and specific products such as used cars, food items like Nutella, and household items like heated beverages. The video focuses on how the cold wave affects everyday life, particularly the usage of items like eggs, Nutella, and used cars, with the intention of introducing a new series titled "열었어" (Heated) series. |
Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula. | [
"floods in the korean peninsula",
"\"Korean Peninsula floods\"",
"floods on the Korean Peninsula during the current day",
"\"Korean Peninsula floods\"",
"\"Floods in South Korea\""
] | [
"floods in korea",
"\"Korean Peninsula floods in June and August\"",
"\"North Korea affected by regional floods during the current rainy season\"",
"\"North Korea flood relief efforts underway amid ongoing regional floods\"",
"\"Floods in Korea\""
] | [
"Heavy rainfall in the southern parts of North Korea in the spring",
"\"Korean Peninsula floods - typhoon or storm, [specific time]\"",
"\"Floods in the Korean Peninsula, South Korea, at [current time]\"",
"\"Korean Peninsula floods in the time of -\""
] | {
"article_url": "https://ko.wikipedia.org/wiki/2020%EB%85%84_%ED%95%9C%EA%B5%AD_%EC%A7%91%EC%A4%91%ED%98%B8%EC%9A%B0",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\n\nBased on the given query, I determine that the following events are likely to be visually represented in a video: \n\nA severe flood, heavy rainfall, and regional rainy season. Considering the query specifically mentions \"North and South Korea\", I infer that the flood should be in the southern parts of the Korean Peninsula, which are prone to heavy rains during this season.\n\nEVENTS:\n\n1. Heavy rainfall in the southern parts of North and South Korea\n2. Floods in North Korea\n3. Floods in South Korea\n4. Regional rainy season in Korea",
"en_article_excerpt": "Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula.",
"en_article_url": "https://en.wikipedia.org/wiki/2020_Korean_floods",
"event_category": "disasters",
"event_name": "2020_korean_floods",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\n\nThis query suggests the event of floods in a specific time frame (between June and August 2020) and a region (North and South Korea). Given the severity of the floods and their impact on the Korean Peninsula, it makes sense to consider events related to natural disasters or extreme weather events that could cause flooding in the same region. This suggests extracting events related to extreme weather or climate-related disasters.\n\nEVENTS:\n\n1. Heavy rainfall\n2. Flooding\n3. Severe weather\n4. Drought\n5. Watergate",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nGiven the query, it is reasonable to extract events that describe or involve flooding. This can include real-world events that occurred between June and August 2020, especially those that involved severe flooding in areas of North and South Korea. Considering the region's climate, these types of events are likely to be relevant and concrete enough to be visualized in a video. Therefore, we can expect to find events that involve floods, which can be described in a sentence or phrase.\n\nEVENTS:\n1. Floods in the far southern parts of the Korean Peninsula\n2. Heavy rainfall in June and August\n3. Regions in North Korea affected by the regional rainy season\n4. Regional floods that severely impacted large areas of both North and South Korea\n5. Floods that resulted from heavy rainfall",
"video_URL": "http://youtube.com/watch?v=SPZkwqJerMU",
"video_description": "2020년 홍수 피해 주민들에 1천483억원 보상 2020년 여름 기나긴 장마로 큰 홍수 피해를 당한 주민들이 총 1천480여억원의 보상금을 지급받게 됐습니다. 환경부 소속 중앙환경분쟁조정위원회는 홍수 피해를 본 주민들과 정부, 지자체 및 한국수자원공사 간 분쟁조정 절차가 최근 마무리됐다고 밝혔습니다. 조정위는 관계부처 원인 조사에서 댐·하천을 미흡하게 관리한 사실이 드러난 점 등을 고려했다고 밝혔습니다. 연합뉴스TV 기사문의 및 제보 : 카톡/라인 jebo23 ▣ 연합뉴스TV 유튜브 채널 구독 https://goo.gl/VuCJMi ▣ 대한민국 뉴스의 시작 연합뉴스TV / Yonhap News TV http://www.yonhapnewstv.co.kr/",
"video_id": "SPZkwqJerMU"
} | 16 | [
"A news anchor is presenting a report in a studio, with a background image of a cityscape at night featuring a brightly lit bridge. The text on the screen indicates that in 2020, 1,483 households affected by flood disasters in South Korea will receive 1,483 billion won in assistance. The broadcast time is 23:27.",
... | SPZkwqJerMU | A devastating flood in South Korea in 2020 affected 1,483 households with 1.483 billion won in financial assistance. The flood led to extensive damage and destruction, but a massive relief effort provided financial compensation to affected residents. Cleanup workers were actively engaged in the recovery process. |
Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula. | [
"floods in the korean peninsula",
"\"Korean Peninsula floods\"",
"floods on the Korean Peninsula during the current day",
"\"Korean Peninsula floods\"",
"\"Floods in South Korea\""
] | [
"floods in korea",
"\"Korean Peninsula floods in June and August\"",
"\"North Korea affected by regional floods during the current rainy season\"",
"\"North Korea flood relief efforts underway amid ongoing regional floods\"",
"\"Floods in Korea\""
] | [
"Heavy rainfall in the southern parts of North Korea in the spring",
"\"Korean Peninsula floods - typhoon or storm, [specific time]\"",
"\"Floods in the Korean Peninsula, South Korea, at [current time]\"",
"\"Korean Peninsula floods in the time of -\""
] | {
"article_url": "https://ko.wikipedia.org/wiki/2020%EB%85%84_%ED%95%9C%EA%B5%AD_%EC%A7%91%EC%A4%91%ED%98%B8%EC%9A%B0",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\n\nBased on the given query, I determine that the following events are likely to be visually represented in a video: \n\nA severe flood, heavy rainfall, and regional rainy season. Considering the query specifically mentions \"North and South Korea\", I infer that the flood should be in the southern parts of the Korean Peninsula, which are prone to heavy rains during this season.\n\nEVENTS:\n\n1. Heavy rainfall in the southern parts of North and South Korea\n2. Floods in North Korea\n3. Floods in South Korea\n4. Regional rainy season in Korea",
"en_article_excerpt": "Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula.",
"en_article_url": "https://en.wikipedia.org/wiki/2020_Korean_floods",
"event_category": "disasters",
"event_name": "2020_korean_floods",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\n\nThis query suggests the event of floods in a specific time frame (between June and August 2020) and a region (North and South Korea). Given the severity of the floods and their impact on the Korean Peninsula, it makes sense to consider events related to natural disasters or extreme weather events that could cause flooding in the same region. This suggests extracting events related to extreme weather or climate-related disasters.\n\nEVENTS:\n\n1. Heavy rainfall\n2. Flooding\n3. Severe weather\n4. Drought\n5. Watergate",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nGiven the query, it is reasonable to extract events that describe or involve flooding. This can include real-world events that occurred between June and August 2020, especially those that involved severe flooding in areas of North and South Korea. Considering the region's climate, these types of events are likely to be relevant and concrete enough to be visualized in a video. Therefore, we can expect to find events that involve floods, which can be described in a sentence or phrase.\n\nEVENTS:\n1. Floods in the far southern parts of the Korean Peninsula\n2. Heavy rainfall in June and August\n3. Regions in North Korea affected by the regional rainy season\n4. Regional floods that severely impacted large areas of both North and South Korea\n5. Floods that resulted from heavy rainfall",
"video_URL": "http://youtube.com/watch?v=mtt7-wXFwxI",
"video_description": "(서울=뉴스1) 김연수 기자 = 지난 1일부터 전국을 강타하고 있는 집중호우로 전국에서 피해가 속출하고 있습니다. 특히 큰 비가 내린 남부 지방을 중심으로 지난 7일 이후 1만1717건의 시설피해가 집계됐다. 섬진강댐 하류 제방이 붕괴돼 주민들이 긴급대피했고 화동군 화개장터가 32년 만에 침수됐습니다. 도로와 교량, 주택 등 1만7879건의 시설피해가 접수된 가운데 제5호 태풍 '장미'가 북상하면서 피해 규모가 더욱 커질 것으로 전망되는데요. 수마가 할퀴고 간 자리 영상으로 확인해보세요. #장마 #홍수피해 #홍수 #물폭탄",
"video_id": "mtt7-wXFwxI"
} | 16 | [
"An aerial view shows a flooded area with submerged buildings and vehicles. The text in Korean indicates a major flood event in a specific location.",
"The image shows a flooded street with water reaching the doors of buildings. The text in Korean translates to \"Local 3,000 households affected,\" suggesting the ... | mtt7-wXFwxI | A major flood event has caused widespread damage and displaced thousands of households in a densely populated area. A rescue operation is underway, with individuals wearing life jackets navigating through flooded streets in a small boat or boat. The floodwaters are rising rapidly, indicating the severity of the situation. The disaster management department is actively involved in the response efforts, with multiple rescue operations taking place. The flooded area spans several buildings, roads, and nearby rivers, highlighting the impact on local infrastructure and residents. The camera continues to capture the aftermath of the flood, showing the extent of the damage and the efforts to mitigate the situation. |
Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula. | [
"floods in the korean peninsula",
"\"Korean Peninsula floods\"",
"floods on the Korean Peninsula during the current day",
"\"Korean Peninsula floods\"",
"\"Floods in South Korea\""
] | [
"floods in korea",
"\"Korean Peninsula floods in June and August\"",
"\"North Korea affected by regional floods during the current rainy season\"",
"\"North Korea flood relief efforts underway amid ongoing regional floods\"",
"\"Floods in Korea\""
] | [
"Heavy rainfall in the southern parts of North Korea in the spring",
"\"Korean Peninsula floods - typhoon or storm, [specific time]\"",
"\"Floods in the Korean Peninsula, South Korea, at [current time]\"",
"\"Korean Peninsula floods in the time of -\""
] | {
"article_url": "https://ko.wikipedia.org/wiki/2020%EB%85%84_%ED%95%9C%EA%B5%AD_%EC%A7%91%EC%A4%91%ED%98%B8%EC%9A%B0",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\n\nBased on the given query, I determine that the following events are likely to be visually represented in a video: \n\nA severe flood, heavy rainfall, and regional rainy season. Considering the query specifically mentions \"North and South Korea\", I infer that the flood should be in the southern parts of the Korean Peninsula, which are prone to heavy rains during this season.\n\nEVENTS:\n\n1. Heavy rainfall in the southern parts of North and South Korea\n2. Floods in North Korea\n3. Floods in South Korea\n4. Regional rainy season in Korea",
"en_article_excerpt": "Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula.",
"en_article_url": "https://en.wikipedia.org/wiki/2020_Korean_floods",
"event_category": "disasters",
"event_name": "2020_korean_floods",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\n\nThis query suggests the event of floods in a specific time frame (between June and August 2020) and a region (North and South Korea). Given the severity of the floods and their impact on the Korean Peninsula, it makes sense to consider events related to natural disasters or extreme weather events that could cause flooding in the same region. This suggests extracting events related to extreme weather or climate-related disasters.\n\nEVENTS:\n\n1. Heavy rainfall\n2. Flooding\n3. Severe weather\n4. Drought\n5. Watergate",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nGiven the query, it is reasonable to extract events that describe or involve flooding. This can include real-world events that occurred between June and August 2020, especially those that involved severe flooding in areas of North and South Korea. Considering the region's climate, these types of events are likely to be relevant and concrete enough to be visualized in a video. Therefore, we can expect to find events that involve floods, which can be described in a sentence or phrase.\n\nEVENTS:\n1. Floods in the far southern parts of the Korean Peninsula\n2. Heavy rainfall in June and August\n3. Regions in North Korea affected by the regional rainy season\n4. Regional floods that severely impacted large areas of both North and South Korea\n5. Floods that resulted from heavy rainfall",
"video_URL": "http://youtube.com/watch?v=5j_NI2eL3HA",
"video_description": "북한에서 집중 호우로 일어난 홍수에 22명이 숨지고 4명이 실종됐습니다. 국제적십자사연맹은 미국의소리 방송에 보낸 이메일에서 \"북한 조선 적십자회와 국가 비상 재해 위원회에 따르면 심각한 홍수로 인해 22명이 사망하고 4명이 실종됐다\"고 밝혔습니다. https://imnews.imbc.com/replay/2020/nwtoday/article/5873504_32531.html #북한 #집중호우 #홍수",
"video_id": "5j_NI2eL3HA"
} | 16 | [
"The image shows a news anchor standing in front of a backdrop depicting a body of water with a pier. The anchor, dressed in a suit and tie, is holding a book or magazine with Korean text. The text on the screen indicates a news broadcast from MBC, showing the time as 7:23. The headline on the screen is in Korean a... | 5j_NI2eL3HA | A severe flood has struck the central region of South Korea, with 22 people reported missing and 4 deaths. The disaster is being managed through rescue efforts and flood management activities, involving a front loader on a construction site in the affected area. The floodwaters are strong and dangerous, causing widespread damage and displacement. The situation remains unpredictable, with the floodwaters continuing to move and affect the area. The disaster is the most severe of its kind, with ongoing efforts to mitigate its impact and provide assistance to those affected. |
Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula. | [
"floods in the korean peninsula",
"\"Korean Peninsula floods\"",
"floods on the Korean Peninsula during the current day",
"\"Korean Peninsula floods\"",
"\"Floods in South Korea\""
] | [
"floods in korea",
"\"Korean Peninsula floods in June and August\"",
"\"North Korea affected by regional floods during the current rainy season\"",
"\"North Korea flood relief efforts underway amid ongoing regional floods\"",
"\"Floods in Korea\""
] | [
"Heavy rainfall in the southern parts of North Korea in the spring",
"\"Korean Peninsula floods - typhoon or storm, [specific time]\"",
"\"Floods in the Korean Peninsula, South Korea, at [current time]\"",
"\"Korean Peninsula floods in the time of -\""
] | {
"article_url": "https://ko.wikipedia.org/wiki/2020%EB%85%84_%ED%95%9C%EA%B5%AD_%EC%A7%91%EC%A4%91%ED%98%B8%EC%9A%B0",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\n\nBased on the given query, I determine that the following events are likely to be visually represented in a video: \n\nA severe flood, heavy rainfall, and regional rainy season. Considering the query specifically mentions \"North and South Korea\", I infer that the flood should be in the southern parts of the Korean Peninsula, which are prone to heavy rains during this season.\n\nEVENTS:\n\n1. Heavy rainfall in the southern parts of North and South Korea\n2. Floods in North Korea\n3. Floods in South Korea\n4. Regional rainy season in Korea",
"en_article_excerpt": "Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula.",
"en_article_url": "https://en.wikipedia.org/wiki/2020_Korean_floods",
"event_category": "disasters",
"event_name": "2020_korean_floods",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\n\nThis query suggests the event of floods in a specific time frame (between June and August 2020) and a region (North and South Korea). Given the severity of the floods and their impact on the Korean Peninsula, it makes sense to consider events related to natural disasters or extreme weather events that could cause flooding in the same region. This suggests extracting events related to extreme weather or climate-related disasters.\n\nEVENTS:\n\n1. Heavy rainfall\n2. Flooding\n3. Severe weather\n4. Drought\n5. Watergate",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nGiven the query, it is reasonable to extract events that describe or involve flooding. This can include real-world events that occurred between June and August 2020, especially those that involved severe flooding in areas of North and South Korea. Considering the region's climate, these types of events are likely to be relevant and concrete enough to be visualized in a video. Therefore, we can expect to find events that involve floods, which can be described in a sentence or phrase.\n\nEVENTS:\n1. Floods in the far southern parts of the Korean Peninsula\n2. Heavy rainfall in June and August\n3. Regions in North Korea affected by the regional rainy season\n4. Regional floods that severely impacted large areas of both North and South Korea\n5. Floods that resulted from heavy rainfall",
"video_URL": "http://youtube.com/watch?v=nR08Eart4vo",
"video_description": "댐 관리를 담당하는 한국수자원공사가 이번 수해와 관련해 댐 수위 조절에는 문제가 없었다고 밝혔습니다. 수자원공사는 댐관리규정 등에 따라 홍수기 기상변화와 댐 상·하류 상황, 댐 안전과 민원 등을 고려해 홍수 대응이 가능하도록 댐 수위를 관리했다고 설명했습니다. 섬진강댐은 집중호우 전부터 홍수기제한수위보다 3m 낮게 유지했고, 용담댐이나 합천댐도 제한 수위를 지켜 전혀 이상이 없었다고 강조했습니다. 다만, 유례없는 최장 기간의 장마로 제방이 무너지고 물이 넘치는 등 복합적인 요인 탓에 홍수 피해가 난 것으로 보인다고 덧붙였습니다. 이에 관련 기관과 함께 원인 조사를 한 뒤 대책 수립에 적극 협조하고, 복구 지원에 노력하겠다고 전했습니다. ※ '당신의 제보가 뉴스가 됩니다' YTN은 여러분의 소중한 제보를 기다립니다. [카카오톡] YTN을 검색해 채널 추가 [전화] 02-398-8585 [메일] social@ytn.co.kr [온라인 제보] www.ytn.co.kr ▶ 기사 원문 : https://www.ytn.co.kr/_ln/0103_202008121704378843 ▶ 제보 안내 : http://goo.gl/gEvsAL, 모바일앱, social@ytn.co.kr, #2424 ▣ YTN 유튜브 채널 구독 : http://goo.gl/Ytb5SZ ⓒ YTN & YTN PLUS 무단 전재 및 재배포금지",
"video_id": "nR08Eart4vo"
} | 16 | [
"A news anchor is presenting a report in front of a large screen displaying the K-water building. The text at the bottom of the screen reads, \"댐, 홍수기 제한수위 이하 유지…피해 원인 조사\" which translates to \"Dam, maintaining below flood season restriction level…investigating cause of damage.\"",
"The news anchor continues to ... | nR08Eart4vo | A severe flooding event in the K-water region has resulted in significant damage to infrastructure, agriculture, and living conditions. The investigation into the cause of the flooding is ongoing, with the primary focus on dam and flood season regulations. A severe flood warning has been issued, with emergency measures in place to prevent further damage. The investigation is focusing on the dams, flood season restrictions, and water discharge capacity, emphasizing the critical nature of this time. The scene highlights the extensive flooding, agricultural damage, and the urgent need for effective measures to control water levels. |
Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula. | [
"floods in the korean peninsula",
"\"Korean Peninsula floods\"",
"floods on the Korean Peninsula during the current day",
"\"Korean Peninsula floods\"",
"\"Floods in South Korea\""
] | [
"floods in korea",
"\"Korean Peninsula floods in June and August\"",
"\"North Korea affected by regional floods during the current rainy season\"",
"\"North Korea flood relief efforts underway amid ongoing regional floods\"",
"\"Floods in Korea\""
] | [
"Heavy rainfall in the southern parts of North Korea in the spring",
"\"Korean Peninsula floods - typhoon or storm, [specific time]\"",
"\"Floods in the Korean Peninsula, South Korea, at [current time]\"",
"\"Korean Peninsula floods in the time of -\""
] | {
"article_url": "https://ko.wikipedia.org/wiki/2020%EB%85%84_%ED%95%9C%EA%B5%AD_%EC%A7%91%EC%A4%91%ED%98%B8%EC%9A%B0",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\n\nBased on the given query, I determine that the following events are likely to be visually represented in a video: \n\nA severe flood, heavy rainfall, and regional rainy season. Considering the query specifically mentions \"North and South Korea\", I infer that the flood should be in the southern parts of the Korean Peninsula, which are prone to heavy rains during this season.\n\nEVENTS:\n\n1. Heavy rainfall in the southern parts of North and South Korea\n2. Floods in North Korea\n3. Floods in South Korea\n4. Regional rainy season in Korea",
"en_article_excerpt": "Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula.",
"en_article_url": "https://en.wikipedia.org/wiki/2020_Korean_floods",
"event_category": "disasters",
"event_name": "2020_korean_floods",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\n\nThis query suggests the event of floods in a specific time frame (between June and August 2020) and a region (North and South Korea). Given the severity of the floods and their impact on the Korean Peninsula, it makes sense to consider events related to natural disasters or extreme weather events that could cause flooding in the same region. This suggests extracting events related to extreme weather or climate-related disasters.\n\nEVENTS:\n\n1. Heavy rainfall\n2. Flooding\n3. Severe weather\n4. Drought\n5. Watergate",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nGiven the query, it is reasonable to extract events that describe or involve flooding. This can include real-world events that occurred between June and August 2020, especially those that involved severe flooding in areas of North and South Korea. Considering the region's climate, these types of events are likely to be relevant and concrete enough to be visualized in a video. Therefore, we can expect to find events that involve floods, which can be described in a sentence or phrase.\n\nEVENTS:\n1. Floods in the far southern parts of the Korean Peninsula\n2. Heavy rainfall in June and August\n3. Regions in North Korea affected by the regional rainy season\n4. Regional floods that severely impacted large areas of both North and South Korea\n5. Floods that resulted from heavy rainfall",
"video_URL": "http://youtube.com/watch?v=GwY9HYVEQIk",
"video_description": "전국 하천 홍수특보 전부 해제…상수도 93곳 피해 중부지방 집중호우가 소강상태에 접어든 가운데 전국 하천 주요 지점의 홍수특보가 전부 해제됐습니다. 환경부는 낙동강이 흐르는 경남 밀양시 삼랑진교에 내려졌던 홍수주의보가 오늘(12일) 오후 5시30분 해제된 것을 끝으로 전국 하천 주요지점에 내려진 홍수특보가 모두 해제됐다고 밝혔습니다. 한강과 임진강 비룡대교·영평교, 왕숙천 진관교 등지에 발령됐던 홍수주의보와 섬진강 하동군 읍내리 지점의 홍수주의보는 각각 지난 11일 오후 6시와 오후 5시 해제됐습니다. 이번 호우로 27개 시·군의 상수도 시설 93곳에서 피해가 발생했습니다. 연합뉴스TV 기사문의 및 제보 : 카톡/라인 jebo23 ▣ 연합뉴스TV 유튜브 채널 구독 https://goo.gl/VuCJMi ▣ 대한민국 뉴스의 시작 연합뉴스TV / Yonhap News TV http://www.yonhapnewstv.co.kr/",
"video_id": "GwY9HYVEQIk"
} | 16 | [
"The image shows two news anchors, a man and a woman, sitting in a news studio. The background displays multiple monitors showing various camera feeds. The news anchors are delivering a news report. The headline at the bottom of the screen reads \"전국 하천 홍수특보 전부 해제...상수도 93곳 피해,\" which translates to \"All flood war... | GwY9HYVEQIk | The video depicts a severe flood situation, with a nationwide flood warning and 93 water supply areas affected. The flood warnings have been lifted, but the water level in rivers continues to rise, and the affected areas still suffer from the aftermath of the flooding. The flood situation is ongoing, with flooded rivers, submerged buildings, and damaged water supply systems. A 5-story building's basement has collapsed, highlighting the severity of the situation. |
Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula. | [
"floods in the korean peninsula",
"\"Korean Peninsula floods\"",
"floods on the Korean Peninsula during the current day",
"\"Korean Peninsula floods\"",
"\"Floods in South Korea\""
] | [
"floods in korea",
"\"Korean Peninsula floods in June and August\"",
"\"North Korea affected by regional floods during the current rainy season\"",
"\"North Korea flood relief efforts underway amid ongoing regional floods\"",
"\"Floods in Korea\""
] | [
"Heavy rainfall in the southern parts of North Korea in the spring",
"\"Korean Peninsula floods - typhoon or storm, [specific time]\"",
"\"Floods in the Korean Peninsula, South Korea, at [current time]\"",
"\"Korean Peninsula floods in the time of -\""
] | {
"article_url": "https://ko.wikipedia.org/wiki/2020%EB%85%84_%ED%95%9C%EA%B5%AD_%EC%A7%91%EC%A4%91%ED%98%B8%EC%9A%B0",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\n\nBased on the given query, I determine that the following events are likely to be visually represented in a video: \n\nA severe flood, heavy rainfall, and regional rainy season. Considering the query specifically mentions \"North and South Korea\", I infer that the flood should be in the southern parts of the Korean Peninsula, which are prone to heavy rains during this season.\n\nEVENTS:\n\n1. Heavy rainfall in the southern parts of North and South Korea\n2. Floods in North Korea\n3. Floods in South Korea\n4. Regional rainy season in Korea",
"en_article_excerpt": "Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula.",
"en_article_url": "https://en.wikipedia.org/wiki/2020_Korean_floods",
"event_category": "disasters",
"event_name": "2020_korean_floods",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\n\nThis query suggests the event of floods in a specific time frame (between June and August 2020) and a region (North and South Korea). Given the severity of the floods and their impact on the Korean Peninsula, it makes sense to consider events related to natural disasters or extreme weather events that could cause flooding in the same region. This suggests extracting events related to extreme weather or climate-related disasters.\n\nEVENTS:\n\n1. Heavy rainfall\n2. Flooding\n3. Severe weather\n4. Drought\n5. Watergate",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nGiven the query, it is reasonable to extract events that describe or involve flooding. This can include real-world events that occurred between June and August 2020, especially those that involved severe flooding in areas of North and South Korea. Considering the region's climate, these types of events are likely to be relevant and concrete enough to be visualized in a video. Therefore, we can expect to find events that involve floods, which can be described in a sentence or phrase.\n\nEVENTS:\n1. Floods in the far southern parts of the Korean Peninsula\n2. Heavy rainfall in June and August\n3. Regions in North Korea affected by the regional rainy season\n4. Regional floods that severely impacted large areas of both North and South Korea\n5. Floods that resulted from heavy rainfall",
"video_URL": "http://youtube.com/watch?v=FG82wp__uZ8",
"video_description": "#홍수 #성금 ●BTN불교TV 유튜브 구독하기 https://bit.ly/3k5nt10 (링크 누르고 구독 버튼 클릭!) ●BTN불교TV는? BTN불교TV(주)는 25년의 시간 동안 부처님의 법음을 널리 홍포해온 한국을 대표하는 불교방송국입니다. 불자님들의 소중한 후원금과 광고수입은 더 나은 방송 포교와 불교문화 진흥, 다양한 컨텐츠를 제작하는데 사용되고 있습니다. BTN은 앞으로도 방송 포교사의 소임을 다하겠습니다. ●BTN불교TV 채널번호 안내 IPTV: KTolleh:233번, SK B TV:305번, LGU+TV:275번(전국공통) SKYLIFE: 181번(전국공통) 지역케이블: 케이블 가입자의 경우 BTN채널번호가 지역별로 달라서 아래 주소를 클릭하셔서 확인하시면 됩니다 https://blog.naver.com/btnii/221391809765 ●다양한 경로로 BTN불교TV를 만나보세요! 홈페이지:http://www.btn.co.kr/index.asp 유튜브:https://www.youtube.com/channel/UCOozqNhXDcb6gg6oaU7uImA?view_as=subscriber 페이스북:https://www.facebook.com/BTNkorea/ 블로그:https://blog.naver.com/btnii 카카오스토리:https://story.kakao.com/ch/btntv",
"video_id": "FG82wp__uZ8"
} | 16 | [
"The image depicts a group of people wearing masks in a crowded public setting, likely to indicate precautions against the spread of COVID-19. The text on the image is in Korean and translates to \"The Republic of Korea, receiving a high rating for its response to COVID-19.\" This suggests that the image is highlig... | FG82wp__uZ8 | The video depicts a disaster in South Korea, specifically a severe rainfall caused by a typhoon in August 2020. The event is shown in a series of frames, highlighting various aspects of the disaster:
- A group of people wearing masks in a crowded public setting is seen in the first frame, indicating precautions against the spread of COVID-19.
- A dramatic sky with lightning and mountains in the background is depicted in the second frame, suggesting a significant event or time period relevant to the video's narrative.
- The natural disaster, described as a disaster caused by unprecedented heavy rainfall, is shown in subsequent frames, emphasizing its severity and impact on the region.
- A news anchor reports on the disaster, stating that the Republic of Korea received a high rating for its response to COVID-19, highlighting the country's effective handling of the pandemic.
- The disaster is further supported by images of a damaged traditional Korean house with a collapsed roof, a news anchor reporting on the damage, and a temple in the background.
- The temple is shown to have sustained significant structural damage, and a large number of people are gathered to repair and clean up the site.
- The community comes together to rebuild and restore the temple, with a group of people using shovels to mix concrete and a person meditating in the background, emphasizing the collective effort and dedication.
- A serene image of a tree with sunlight filtering through its leaves is shown in the final frame, highlighting a moment of reflection and contemplation amidst the chaos.
- The final frame features a silhouette of a person meditating against a backdrop of a sunrise or sunset, with contact information for a fundraising effort related to disaster relief and spiritual well-being. |
Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula. | [
"floods in the korean peninsula",
"\"Korean Peninsula floods\"",
"floods on the Korean Peninsula during the current day",
"\"Korean Peninsula floods\"",
"\"Floods in South Korea\""
] | [
"floods in korea",
"\"Korean Peninsula floods in June and August\"",
"\"North Korea affected by regional floods during the current rainy season\"",
"\"North Korea flood relief efforts underway amid ongoing regional floods\"",
"\"Floods in Korea\""
] | [
"Heavy rainfall in the southern parts of North Korea in the spring",
"\"Korean Peninsula floods - typhoon or storm, [specific time]\"",
"\"Floods in the Korean Peninsula, South Korea, at [current time]\"",
"\"Korean Peninsula floods in the time of -\""
] | {
"article_url": "https://ko.wikipedia.org/wiki/2020%EB%85%84_%ED%95%9C%EA%B5%AD_%EC%A7%91%EC%A4%91%ED%98%B8%EC%9A%B0",
"during_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to decompose the ongoing event in multiple simple events. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"during_raw_generation": "EXPLANATION:\n\nBased on the given query, I determine that the following events are likely to be visually represented in a video: \n\nA severe flood, heavy rainfall, and regional rainy season. Considering the query specifically mentions \"North and South Korea\", I infer that the flood should be in the southern parts of the Korean Peninsula, which are prone to heavy rains during this season.\n\nEVENTS:\n\n1. Heavy rainfall in the southern parts of North and South Korea\n2. Floods in North Korea\n3. Floods in South Korea\n4. Regional rainy season in Korea",
"en_article_excerpt": "Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula.",
"en_article_url": "https://en.wikipedia.org/wiki/2020_Korean_floods",
"event_category": "disasters",
"event_name": "2020_korean_floods",
"language": "korean",
"prequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what previous events could lead to this given event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"prequel_raw_generation": "EXPLANATION:\n\nThis query suggests the event of floods in a specific time frame (between June and August 2020) and a region (North and South Korea). Given the severity of the floods and their impact on the Korean Peninsula, it makes sense to consider events related to natural disasters or extreme weather events that could cause flooding in the same region. This suggests extracting events related to extreme weather or climate-related disasters.\n\nEVENTS:\n\n1. Heavy rainfall\n2. Flooding\n3. Severe weather\n4. Drought\n5. Watergate",
"sequel_prompt": "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nCutting Knowledge Date: December 2023\nToday Date: 16 Jan 2025\n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nYou are given a video search query. Assume the video search query is an ongoing event. You have to find out what events can be the outcome of this ongoing event. The extracted events should be concrete enough to be visualized in a video.\n\nHere are some guidelines: \n- Extract events that could be visualized in a video.\n- Before extracting an event, consider whether the event text could serve as a reasonable caption for a frame of a video associated with the query.\n- Give specific events related to the query, use your previous WORLD KNOWLEDGE to extract events that are likely to be seen in a video. For example, if the query is \"2019 Christmas Eve Protests in Hong Kong\", you might return \"Protesters marching in the streets\", but not \"Protestors marching in snow\" because it is not likely to have snow in Hong Kong, though its pretty common in other places. \n- First, explain your reasoning in 1-2 sentences about which events it makes sense to extract. Then, provide a numbered list with the events (anywhere from one to five event; use your best judgement to determine how many to extract)\n- Structure your response into two sections, \"\"EXPLANATION:\"\" and then \"\"EVENTS:\"\". The events sections should contain nothing but the events. \n\nHere is the query I'd like you to extract events from: Between June and August 2020, floods severely impacted large areas of both North and South Korea due to heavy rains of the regional rainy season, primarily in the far southern parts of the Korean Peninsula.<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
"sequel_raw_generation": "EXPLANATION:\nGiven the query, it is reasonable to extract events that describe or involve flooding. This can include real-world events that occurred between June and August 2020, especially those that involved severe flooding in areas of North and South Korea. Considering the region's climate, these types of events are likely to be relevant and concrete enough to be visualized in a video. Therefore, we can expect to find events that involve floods, which can be described in a sentence or phrase.\n\nEVENTS:\n1. Floods in the far southern parts of the Korean Peninsula\n2. Heavy rainfall in June and August\n3. Regions in North Korea affected by the regional rainy season\n4. Regional floods that severely impacted large areas of both North and South Korea\n5. Floods that resulted from heavy rainfall",
"video_URL": "http://twitter.com/i/status/1291976724456992776",
"video_description": "섬진강이 난리네요 ㅠ#섬진강#유등면#섬진강 홍수",
"video_id": "1291976724456992776"
} | 16 | [
"The image depicts a flooded area where water has inundated a structure with a blue roof. The water level is high, reaching up to the base of the roof, and several cows are seen standing in the water inside the structure. There are green bushes in the foreground, and some debris is floating on the water surface. Th... | 1291976724456992776 | The video depicts a video of flooding in an agricultural setting, with a blue-roofed structure being submerged by water. The floodwaters have been persistent for several frames, with cows standing in the water and green bushes partially submerged in the foreground. The situation remains unchanged, with blue debris floating on the surface, indicating a long-standing flooding event. |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.