How vishal0205/google-photo-review-analyser- works
This repository analyses Google Photos Play Store reviews through two complementary stages: notebook-based LLM classification and evaluation, followed by a Streamlit app for exploring labelled results. In the app, dashboard code loads the labelled CSV through analytics helpers and presents summaries, filters, trends, complaints and review details. Its chatbot retrieves relevant reviews and answers questions, using Gemini for descriptive responses and exact dataset statistics for numerical questions. The notebook and classification workflow are documented but not excerpted, so their internal calls are not mapped.
Main entry points: Review_Analyser_Starter.ipynb Classification notebook · dashboard.py Streamlit dashboard · labelled_reviews.csv Labelled reviews · chatbot.py Review chatbot
Components 9
Classification and evaluation
- Classification notebook
Review_Analyser_Starter.ipynb - Manual labels
manual_labels_50.csv
Review data
- Source reviews
google_photos_reviews.csv - Labelled reviews
labelled_reviews.csv
Dashboard analytics
- Streamlit dashboard
streamlit_app/dashboard.py - Review analytics
streamlit_app/analytics.py
AI assistant
- Review chatbot
streamlit_app/chatbot.py
Other components
- Dashboard user
- Gemini API
Connections 8
- Source reviews → Classification notebook: analyses
README.md - Classification notebook → Labelled reviews: classifies
README.md - Classification notebook → Manual labels: evaluates against
README.md - Labelled reviews → Review analytics: loads
streamlit_app/analytics.py - Streamlit dashboard → Review analytics: calls
streamlit_app/dashboard.py - Dashboard user → Streamlit dashboard: explores
README.md - Streamlit dashboard → Review chatbot: uses
README.md - Review chatbot → Gemini API: generates answers
README.md