license: cc-by-4.0
task_categories:
- text-classification
- text-generation
language:
- en
tags:
- sentiment-analysis
- education
- generative-ai
- app-reviews
dataset_info:
features:
- name: app_name
dtype: string
- name: source_file
dtype: string
- name: review_reviewId
dtype: string
- name: review_userName
dtype: string
- name: review_userImage
dtype: string
- name: review_content
dtype: string
- name: review_score
dtype: int64
- name: review_thumbsUpCount
dtype: int64
- name: review_reviewCreatedVersion
dtype: string
- name: review_at
dtype: string
- name: review_replyContent
dtype: string
- name: review_repliedAt
dtype: string
- name: review_appVersion
dtype: string
- name: app_info_json
dtype: string
- name: app_title
dtype: string
- name: app_summary
dtype: string
- name: app_description
dtype: string
- name: app_descriptionHTML
dtype: string
- name: app_genre
dtype: string
- name: app_genre_id
dtype: string
- name: app_developer
dtype: string
- name: app_developer_email
dtype: string
- name: app_installs
dtype: string
- name: app_minInstalls
dtype: int64
- name: app_realInstalls
dtype: int64
- name: app_score
dtype: float64
- name: app_ratings
dtype: int64
- name: app_reviews_count
dtype: int64
- name: app_price
dtype: int64
- name: app_free
dtype: bool
- name: app_offersIAP
dtype: bool
- name: app_inAppProductPrice
dtype: string
- name: app_appId
dtype: string
- name: app_url
dtype: string
- name: app_icon
dtype: string
- name: app_headerImage
dtype: string
- name: app_released
dtype: string
- name: app_lastUpdatedOn
dtype: string
- name: app_updated_ts
dtype: int64
- name: app_version
dtype: string
splits:
- name: train
num_bytes: 7755072015
num_examples: 480831
- name: app_AI_Quiz_Generator
num_bytes: 1758004
num_examples: 109
- name: app_Edu_AI_AI_Homework_Helper
num_bytes: 403211
num_examples: 25
- name: app_Help_AI_Your_Homework_With_AI
num_bytes: 30127996
num_examples: 1868
- name: app_Answer_AI_Your_AI_tutor
num_bytes: 263748994
num_examples: 16353
- name: app_Quiz_AI_AI_Homework_Helper
num_bytes: 4306303
num_examples: 267
- name: app_Teacher_AI_Language_Practice
num_bytes: 887066
num_examples: 55
- name: app_Studocu_AI_Homework_Helper
num_bytes: 37611609
num_examples: 2332
- name: app_Kahoot_Play_Create_Quizzes
num_bytes: 550739124
num_examples: 34147
- name: app_Quizard_AI_Homework_Helper
num_bytes: 23289521
num_examples: 1444
- name: app_Homework_AI_Math_Essay_App
num_bytes: 68594415
num_examples: 4253
- name: app_Course_Hero_AI_Homework_Help
num_bytes: 68223460
num_examples: 4230
- name: app_Academi_AI_Study_Exam_Tutor
num_bytes: 3935348
num_examples: 244
- name: app_Nerd_AI_Tutor_Math_Helper
num_bytes: 237233777
num_examples: 14709
- name: app_Tutor_AI_Learning_Assistants
num_bytes: 193541
num_examples: 12
- name: app_StudyX_AI_Homework_Helper
num_bytes: 98738539
num_examples: 6122
- name: app_School_Hack
num_bytes: 84755149
num_examples: 5255
- name: app_Gauth_AI_Study_Companion
num_bytes: 1672716669
num_examples: 103712
- name: app_Question_AI_Chatbot_Math_AI
num_bytes: 528610853
num_examples: 32775
- name: app_Brainly_AI_Homework_Helper
num_bytes: 3767612368
num_examples: 233600
- name: app_AI_Questions_Generator
num_bytes: 5548196
num_examples: 344
- name: app_Blackboard
num_bytes: 301376732
num_examples: 18686
- name: app_Mathos_AI_Math_Helper_Tutor
num_bytes: 4661130
num_examples: 289
download_size: 196210355
dataset_size: 15510144020
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: app_AI_Quiz_Generator
path: data/app_AI_Quiz_Generator-*
- split: app_Edu_AI_AI_Homework_Helper
path: data/app_Edu_AI_AI_Homework_Helper-*
- split: app_Help_AI_Your_Homework_With_AI
path: data/app_Help_AI_Your_Homework_With_AI-*
- split: app_Answer_AI_Your_AI_tutor
path: data/app_Answer_AI_Your_AI_tutor-*
- split: app_Quiz_AI_AI_Homework_Helper
path: data/app_Quiz_AI_AI_Homework_Helper-*
- split: app_Teacher_AI_Language_Practice
path: data/app_Teacher_AI_Language_Practice-*
- split: app_Studocu_AI_Homework_Helper
path: data/app_Studocu_AI_Homework_Helper-*
- split: app_Kahoot_Play_Create_Quizzes
path: data/app_Kahoot_Play_Create_Quizzes-*
- split: app_Quizard_AI_Homework_Helper
path: data/app_Quizard_AI_Homework_Helper-*
- split: app_Homework_AI_Math_Essay_App
path: data/app_Homework_AI_Math_Essay_App-*
- split: app_Course_Hero_AI_Homework_Help
path: data/app_Course_Hero_AI_Homework_Help-*
- split: app_Academi_AI_Study_Exam_Tutor
path: data/app_Academi_AI_Study_Exam_Tutor-*
- split: app_Nerd_AI_Tutor_Math_Helper
path: data/app_Nerd_AI_Tutor_Math_Helper-*
- split: app_Tutor_AI_Learning_Assistants
path: data/app_Tutor_AI_Learning_Assistants-*
- split: app_StudyX_AI_Homework_Helper
path: data/app_StudyX_AI_Homework_Helper-*
- split: app_School_Hack
path: data/app_School_Hack-*
- split: app_Gauth_AI_Study_Companion
path: data/app_Gauth_AI_Study_Companion-*
- split: app_Question_AI_Chatbot_Math_AI
path: data/app_Question_AI_Chatbot_Math_AI-*
- split: app_Brainly_AI_Homework_Helper
path: data/app_Brainly_AI_Homework_Helper-*
- split: app_AI_Questions_Generator
path: data/app_AI_Questions_Generator-*
- split: app_Blackboard
path: data/app_Blackboard-*
- split: app_Mathos_AI_Math_Helper_Tutor
path: data/app_Mathos_AI_Math_Helper_Tutor-*
🎓 GenAI-EdSent: Sentiment-Driven Evaluation of AI Educational Apps
Unveiling User Perceptions in the Generative AI Era Through Large-Scale Sentiment Analysis
📖 Overview
The integration of Generative AI into education has sparked a digital transformation in e-teaching, yet user perceptions of AI educational apps remain critically underexplored. This research project bridges that gap through a comprehensive sentiment-driven evaluation of user reviews from top AI educational apps on the Google Play Store.
This dataset accompanies the paper Unveiling User Perceptions in the Generative AI Era: A Sentiment-Driven Evaluation of AI Educational Apps' Role in Digital Transformation of E-Teaching.
Code: https://github.com/erfan-nourbakhsh/GenAI-EdSent
🎯 Research Objectives
- Quantify sentiment distributions and distill key positive/negative themes across app categories
- Compare performance trends among different app types (homework helpers, math solvers, LMS, etc.)
- Propose future directions for AI educational ecosystems with hybrid AI-human models
📄 Associated Paper
Title: Unveiling User Perceptions in the Generative AI Era: A Sentiment-Driven Evaluation of AI Educational Apps' Role in Digital Transformation of E-Teaching
Authors: Adeleh Mazaheriyan (Islamic Azad University) & Erfan Nourbakhsh (University of Isfahan)
Abstract: This study performs a sentiment-driven evaluation of user reviews from 22 top AI ed-apps on the Google Play Store to assess efficacy, challenges, and pedagogical implications. Our pipeline leverages RoBERTa for binary sentiment classification, GPT-4o for key point extraction, and GPT-5 for synthesizing top positive/negative themes. Results reveal predominantly positive sentiments, with homework apps leading (e.g., Edu AI: 95.9% positive) while specialized LMS/language apps lag due to stability issues.
📊 Research at a Glance
| 📱 Apps Analyzed | 💬 Reviews Processed | 🤖 AI Models Used | 📚 Categories |
|---|---|---|---|
| 22 | 481,000+ | 3 (RoBERTa, GPT-4o, GPT-5) | 7 |
🔥 Quick Stats
🎯 96.0% positive sentiment for top performer (Edu AI) ⚡ 481K training reviews • 481K validation samples 📈 4.43M total ratings across all apps 🏆 85% average positive sentiment for homework helpers ⚠️ 21.8% positive sentiment for lowest performer (Teacher AI)
🔍 Key Findings
📊 Sentiment Distribution Highlights
Homework Helpers dominate with ~85% positive sentiment
- 🥇 Edu AI: 96.0% positive (accuracy, speed, personalization)
- 🥈 Answer.AI: 92.7% positive (24/7 tutoring, step-by-step explanations)
Math-Focused Solvers show strong performance (~80% positive)
- Users praise problem-solving efficiency and photo recognition
Language/LMS Apps lag behind (20-40% positive)
- ⚠️ Teacher AI: 21.8% positive (instability, limited features)
- Issues: crashes, paywalls, feature gaps
🌟 Top Positive Themes
- ⚡ Efficiency & Speed - Quick solutions for homework and brainstorming
- 🎯 Personalized Learning - Step-by-step explanations tailored to student needs
- 🎮 Engagement - Gamification and community features boost motivation
- 🌍 Accessibility - Democratizing education for under-resourced areas
- 🤝 Multi-Subject Support - Versatile tools across STEM and humanities
⚠️ Top Negative Themes
- 💰 Aggressive Monetization - Restrictive paywalls limiting free features
- ❌ Inaccuracies - Wrong answers eroding trust, especially in specialized domains
- 📺 Excessive Ads - Disrupting learning flow
- 🐛 Technical Glitches - Crashes, slow loading, notation failures
- ⚖️ Equity Concerns - Digital divide and over-reliance risks
🛠️ Methodology
Our systematic pipeline combines web scraping, transformer-based NLP, and large language models to analyze authentic user feedback at scale.
Figure 1: Multi-stage workflow from data collection to theme synthesis
🔄 Pipeline Stages
1️⃣ App Selection & Categorization
- Selected 22 prominent AI education apps based on ratings, downloads, and GenAI integration
- Categorized into 7 functional types with overlaps for multifunctional designs:
- 📝 AI Quiz & Question Generators
- 🎒 All-in-One Study Companions
- ✏️ Homework Helpers
- 🔢 Math-Focused Solvers & Specialized Tools
- 📄 Document/Content Tools
- 🏫 Learning Management Systems (LMS)
- 🌐 Language Learning Apps
2️⃣ Data Collection
- Web scraping from Google Play Store using Python
- Collected app metadata + verbatim user reviews (up to November 2025)
- Dataset: Tens of thousands to millions of reviews per app
3️⃣ Sentiment Analysis & Theme Extraction
| Stage | Model | Purpose |
|---|---|---|
| Binary Classification | RoBERTa | Transformer-based sentiment labeling (positive/negative) |
| Key Point Extraction | GPT-4o | Distill recurring themes and pain points |
| Theme Synthesis | GPT-5 | Generate top 5 positive/negative summaries |
4️⃣ Aggregation & Trend Analysis
- Computed sentiment percentages per app and category
- Enabled cross-app performance comparisons
📈 Results
Figure 2: Positive vs. negative sentiment percentages across 22 AI educational apps
🏆 Top Performers
| Rank | App | Category | Positive % |
|---|---|---|---|
| 1 | Edu AI | Homework Helper | 96.0% |
| 2 | Answer.AI | Multi-Tool | 92.7% |
| 3 | Question.AI | Chatbot/Math | 87.9% |
| 4 | Homework AI | Math/Essay | 85.6% |
| 5 | Help AI | Homework Helper | 85.2% |
⚠️ Areas for Improvement
| Rank | App | Category | Positive % | Key Issues |
|---|---|---|---|---|
| 22 | Teacher AI | Language | 21.8% | Instability, limited features |
| 21 | Tutor AI | Learning Assistant | 41.7% | Narrow functionality |
| 20 | Blackboard | LMS | 53.3% | Technical glitches |
📊 Category-Level Insights
- Homework Helpers: Average 85% positive → Strong personalization and speed
- Math Solvers: Average 80% positive → Photo recognition praised, notation failures criticized
- LMS/Language Apps: Average 35% positive → Urgent need for stability improvements
💡 Future Directions
🔮 Proposed Innovations
Hybrid AI-Human Models 🤝
- Combine app strengths (real-time assistance) with teacher oversight
- Mitigate risks like inaccuracies and over-dependency
VR/AR Integration 🥽
- Immersive learning experiences (virtual labs, interactive simulations)
- Address gaps in engagement and multimodal inputs
Ethical AI Roadmap ⚖️
- For Developers: Adaptive personalization, plagiarism detection, equitable monetization
- For Policymakers: Mandated free tiers, accuracy standards, data privacy protections
📊 Dataset
The full dataset containing 481,000+ app reviews from 22 AI educational apps is publicly available on Hugging Face:
🔗 GenAI-EdSent Dataset on Hugging Face
The dataset includes:
- App Information: Metadata for all 22 applications (ratings, descriptions, install counts)
- User Reviews: Complete review texts with scores, timestamps, and sentiment labels
- Sentiment Analysis Results: Binary classification outputs from RoBERTa
- Extracted Insights: Key positive/negative themes per application
📁 Repository Data Files
This repository includes curated analysis outputs:
1. App_Ratings.json - Application Metadata
Contains Google Play Store ratings and summaries for all 22 apps:
📋 View Sample Data
{
"name": "Edu AI - AI Homework Helper",
"summary": "AI Homework Helper - Math, Physics, Chemical, etc.",
"app_score": 4.55
},
{
"name": "Answer.AI - Your AI tutor",
"summary": "Scan and Get Instant Answers on Your Phone",
"app_score": 4.73
}
Top Rated Apps:
- 🥇 Studocu: 4.86/5 ⭐
- 🥈 Gauth: 4.77/5 ⭐
- 🥉 Kahoot!: 4.74/5 ⭐
Lowest Rated Apps:
- ⚠️ Teacher AI: 2.29/5 ⭐
- ⚠️ Blackboard: 3.39/5 ⭐
- ⚠️ Tutor AI: 3.40/5 ⭐
2. Top_5_Points.json - Sentiment Themes Per App
Contains the top 5 positive and negative points extracted for each application using GPT-4o and GPT-5:
📋 View Sample Analysis (Edu AI)
Positive Points:
- ✅ The app enables users to complete homework tasks effectively
- ✅ The app delivers perfect-quality assignment outputs that meet high standards
Negative Points:
- ❌ The app returns answers in Spanish instead of matching the user's preferred language
- ❌ Text is displayed in a font size that is too small to read comfortably
📋 View Sample Analysis (Answer.AI)
Positive Points:
- ✅ Reliably helps students complete homework across subjects and grade levels
- ✅ Consistently delivers accurate, correct answers
- ✅ Explains solutions step by step with reasoning
- ✅ Provides answers very quickly with fast response times
- ✅ Easy to use with intuitive interface and camera/scan input
Negative Points:
- ❌ Frequently provides incorrect or incomplete answers (especially math/graphs)
- ❌ Core features locked behind expensive paywall
- ❌ Restrictive energy/points system limits free usage
- ❌ App instability with crashes and broken features
- ❌ Unreliable camera/scan and image recognition
Key Insights from Theme Analysis:
- Most Praised: Homework efficiency, step-by-step explanations, speed
- Most Criticized: Paywalls, accuracy issues, technical instability
- Category Trends: Homework helpers excel; LMS/language apps struggle
🚀 Installation
Prerequisites
Python 3.8+
pip
Setup
# Clone the repository
git clone https://github.com/erfan-nourbakhsh/GenAI-EdSent.git
cd GenAI-EdSent
# Install dependencies
pip install -r requirements.txt
Required Libraries
transformers(RoBERTa model)openai(GPT API access)beautifulsoup4/selenium(web scraping)pandas,numpy(data processing)matplotlib,seaborn(visualization)
🔬 Sample Usage
1. Sentiment Classification
# Run RoBERTa-based sentiment analysis
python Positive_Negative_Points_Classifier.py
2. Extract Key Points
# Use GPT-4o to extract themes from reviews
python Positive_Negative_Points_Generator.py
3. Generate Top Themes
# Synthesize top 5 positive/negative points with GPT-5
python Top_5_Points_Generator.py
📚 Citation
If you use this work in your research, please cite:
@article{mazaherian2025unveiling,
title={Unveiling User Perceptions in the Generative AI Era: A Sentiment-Driven Evaluation of AI Educational Apps' Role in Digital Transformation of e-Teaching},
author={Mazaherian, Adeleh and Nourbakhsh, Erfan},
journal={arXiv preprint arXiv:2512.11934},
year={2025}
}
👥 Meet the Researchers
🌟 Behind the Research
A collaborative effort bridging educational theory and AI technology to understand how generative AI is reshaping learning experiences worldwide.
🎓 Adeleh MazaheriyanEducation Researcher
Department of Education Specializes in pedagogical evaluation and digital transformation in e-teaching, bringing educational theory perspective to AI assessment. |
🤖 Erfan NourbakhshAI Researcher & Developer
Artificial Intelligence Department Expert in NLP, sentiment analysis, and LLM applications, developing the technical pipeline for large-scale review analysis. |
🤝 Interdisciplinary Collaboration
This research exemplifies the power of interdisciplinary collaboration, combining:
- 📚 Educational Theory → Understanding pedagogical implications
- 🧠 AI/ML Technology → Analyzing 960K+ reviews at scale
- 📊 Data Science → Extracting actionable insights for stakeholders
Together, we're working to ensure AI in education serves all learners equitably.
🤝 Contributing
Contributions are welcome! Please feel free to submit a Pull Request. For major changes, please open an issue first to discuss proposed changes.
📄 License
This dataset is licensed under the CC-BY-4.0 License. The code is licensed under the MIT License.
🌟 Acknowledgments
We extend our gratitude to:
- Microsoft Education for the 2025 AI in Education Report
- RAND Corporation for K-12 AI adoption insights and teacher engagement studies
- OpenAI & Hugging Face for providing access to state-of-the-art language models (GPT-4o, GPT-5, RoBERTa)
- Google Play Store developers and the educational AI community
- Thousands of users whose reviews made this research possible
📞 Contact & Collaboration
We welcome questions, feedback, and collaboration opportunities!
For Research Inquiries:
- 📧 Email: [email protected] or [email protected]
- 🤗 Dataset: Hugging Face
Interested in:
- Collaborative research on AI in education?
- Extending this analysis to other platforms (Apple App Store, web apps)?
- Developing predictive models for app success?
- Contributing to the codebase?
We'd love to hear from you! 💬
