How candyhuang940730/cinerecommend works
CineRecommend is a Streamlit movie recommender. A user selects or creates a profile, rates movies, then requests genre-filtered recommendations with a slider balancing multimodal content similarity against trust-weighted social ratings. The repository also includes MovieLens loading, TMDB metadata retrieval, feature embedding, trust-score computation, spam injection for experiments, and recommendation evaluation. The graph distinguishes these domain stages and the database access layer. Edges follow excerpts, SOURCE INDEX references, or README statements; unsampled schema details and unshown runtime wiring are not inferred.
Main entry points: db.py Database connection · app.py Streamlit interface · recommend.py Hybrid recommender · fetch_tmdb.py TMDB metadata fetch
Components 14
User experience
- Streamlit interface
app.py - User profiles and ratings
core/profile.py
Recommendation
- Hybrid recommender
core/recommend.py - Text and poster embeddings
features/embed.py - User trust scores
core/trust.py
Data and trust
- MovieLens data
- MovieLens import
etl/load_movielens.py - TMDB metadata fetch
etl/fetch_tmdb.py - Database connection
core/db.py - PostgreSQL and pgvector
schema.sql
Evaluation
- Ranking evaluation
core/evaluate.py - Spam experiment data
etl/inject_spam.py
Other components
- Moviegoer
- TMDB metadata API
Connections 16
- Moviegoer → Streamlit interface: uses
README.md - Streamlit interface → User profiles and ratings: reads and rates
app.py - Streamlit interface → Hybrid recommender: requests recommendations
app.py - Streamlit interface → Database connection: reads year range
app.py - User profiles and ratings → Database connection: reads and writes
core/profile.py - Hybrid recommender → Database connection: queries ratings and movies
core/recommend.py - Ranking evaluation → Hybrid recommender: evaluates rankings
core/evaluate.py - Ranking evaluation → Database connection: reads ratings
core/evaluate.py - User trust scores → Database connection: reads and stores scores
core/trust.py - MovieLens import → Database connection: loads records
etl/load_movielens.py - MovieLens data → MovieLens import: supplies dataset
README.md - TMDB metadata fetch → TMDB metadata API: fetches metadata
etl/fetch_tmdb.py - TMDB metadata fetch → Database connection: updates movie metadata
etl/fetch_tmdb.py - Text and poster embeddings → Database connection: stores vectors
features/embed.py - Spam experiment data → Database connection: inserts synthetic ratings
etl/inject_spam.py - Database connection → PostgreSQL and pgvector: connects to
README.md