How subhamgupta9/cinematch-ai works
CineMatch AI is a Streamlit movie-recommendation and global-discovery application. Users navigate to discovery, search by taste or a movie, browse by genre and region or ratings, and receive recommendations or movie details. The documented hybrid path combines local content and collaborative signals with global TMDb discovery; details include cast, trailers and India streaming options. The graph preserves the documented watchlist and history capabilities. Some subsystem links reflect README-level architecture because their implementation was not sampled.
Main entry points: app.py Streamlit app · hybrid_recommender.py Hybrid recommender · tmdb_client.py Movie metadata client · page_discover.py Discover studio
Components 17
User experience
- Streamlit app
app.py - Navigation and state
ui/components.py - Discover studio
ui/page_discover.py - Movie details
ui/modal_movie_details.py - Accessible theme
ui/styles.py
Recommendation and discovery
- Hybrid recommender
hybrid_recommender.py - Candidate generation
candidate_generator.py
Catalog and metadata
- Global search
tmdb_search.py - Movie metadata client
tmdb_client.py - Unified movie records
unified_data_model.py - MovieLens catalog and ratings
data/ml-latest-small
Recommendation models
- Content similarity
recommender.py - Collaborative filtering
collaborative_recommender.py
Other components
- Viewer
- TMDb API
- JustWatch availability
- YouTube trailers
Connections 21
- Viewer → Streamlit app: interacts with
README.md - Streamlit app → Navigation and state: renders navigation
app.py - Streamlit app → Discover studio: routes to discovery
app.py - Streamlit app → Movie details: renders details
app.py - Streamlit app → Accessible theme: applies styles
README.md - Discover studio → Hybrid recommender: requests recommendations
README.md - Discover studio → Global search: searches titles
ui/page_discover.py - Discover studio → Movie metadata client: fetches discovery results
app.py - Hybrid recommender → Candidate generation: builds candidate pool
hybrid_recommender.py - Hybrid recommender → Content similarity: scores content
README.md - Hybrid recommender → Collaborative filtering: uses affinity
README.md - Hybrid recommender → Global search: adds global candidates
README.md - Candidate generation → Unified movie records: normalizes movies
candidate_generator.py - Candidate generation → MovieLens catalog and ratings: reads catalog data
- Collaborative filtering → MovieLens catalog and ratings: fits rating data
- Global search → TMDb API: searches catalog
README.md - Movie metadata client → TMDb API: fetches movie details
README.md - Movie metadata client → JustWatch availability: gets watch providers
README.md - Movie metadata client → YouTube trailers: provides trailer links
README.md - Movie details → Movie metadata client: loads rich details
ui/modal_movie_details.py - Navigation and state → Streamlit app: tracks navigation state
app.py