How aethermaker/nlp-movie-recommender-streamlit works
This is a Streamlit mood-based movie recommender. A user enters a mood or plot description; the app loads movie data and a sentence model, requests similarity-ranked recommendations, then displays movie details and fetches posters from TMDB. The sampled code shows the recommendation and poster-fetching flow; the README documents the precomputed embeddings and metadata. External library internals are not represented.
Main entry points: app.py Streamlit app · recommender_logic.py Query matching · movie_embeddings.npy Movie embeddings · movie_metadata.parquet Movie metadata
Components 11
User interface
- Streamlit app
app.py
Recommendation
- Query matching
recommender_logic.py - Model loading
recommender_logic.py
Movie data
- Data loading
recommender_logic.py - Movie embeddings
movie_embeddings.npy - Movie metadata
movie_metadata.parquet
External services
- Poster details
recommender_logic.py - Sentence Transformers
- Cosine similarity
- TMDB API
Other components
- Movie seeker
Connections 12
- Movie seeker → Streamlit app: enters query
README.md - Streamlit app → Model loading: loads model
app.py - Streamlit app → Data loading: loads data
app.py - Data loading → Movie embeddings: reads vectors
recommender_logic.py - Data loading → Movie metadata: reads metadata
recommender_logic.py - Streamlit app → Query matching: requests matches
app.py - Query matching → Sentence Transformers: encodes query
recommender_logic.py - Model loading → Sentence Transformers: loads model
recommender_logic.py - Query matching → Cosine similarity: scores vectors
recommender_logic.py - Streamlit app → Poster details: fetches details
app.py - Poster details → TMDB API: requests details
recommender_logic.py - Streamlit app → Movie seeker: displays results
app.py