How akhilsinghkoteri/movie-recommender works
This is a small Streamlit content-based movie recommender. The workflow loads and cleans the movie catalog, builds genre TF-IDF vectors, accepts favorite titles, forms a taste profile, ranks similar movies, and displays the top-N suggestions. The graph separates the interface, recommendation stages, and catalog. The sampled source shows the main execution paths; recommendations are computed on demand rather than through a precomputed similarity matrix.
Main entry points: recommender.py Similarity ranking · app.py Streamlit interface · movies.csv MovieLens catalog
Components 10
Web interface
- Streamlit interface
app.py - Cached engine setup
app.py - Recommendation display
app.py
Recommendation engine
- Genre TF-IDF vectors
recommender.py - Taste profile
recommender.py - Similarity ranking
recommender.py - Top-N suggestions
recommender.py
Movie catalog
- Load and clean data
recommender.py - MovieLens catalog
movies.csv
Other components
- Movie seeker
Connections 13
- Movie seeker → Streamlit interface: selects favorites
app.py - Streamlit interface → Cached engine setup: initializes engine
app.py - Cached engine setup → Load and clean data: loads data
app.py - Load and clean data → MovieLens catalog: reads CSV
recommender.py - Cached engine setup → Genre TF-IDF vectors: builds vectors
app.py - Load and clean data → Genre TF-IDF vectors: provides cleaned genres
recommender.py - Streamlit interface → Similarity ranking: requests recommendations
app.py - Similarity ranking → Taste profile: averages favorite vectors
recommender.py - Taste profile → Similarity ranking: provides taste vector
recommender.py - Similarity ranking → Genre TF-IDF vectors: compares against vectors
recommender.py - Similarity ranking → Top-N suggestions: selects top matches
recommender.py - Top-N suggestions → Recommendation display: returns titles
app.py - Recommendation display → Movie seeker: displays suggestions
app.py