How awantigiradkar/movie-recommendation-system works
Project type and purpose
• This is a lightweight full-stack (or more accurately back-end plus UI) content-based movie recommender system built in Python with a Streamlit interface.
• The system reads movie metadata, computes a cosine-similarity matrix, serves recommendations via an interactive web UI, and fetches movie posters from the TMDb API.
File structure
Top-level files and their roles:
• app.py – main Streamlit application (serves UI and handles user interactions)
• Movie_Recommender_System.ipynb – exploratory notebook (data processing and modeling code)
• tmdb_5000_movies.csv, tmdb_5000_credits.csv – raw dataset sources
• movie_list.pkl, movie_dict.pkl, similarity.pkl – serialized model artifacts (pickles for lookup list, metadata dict, similarity matrix)
• README.md, .gitattributes – documentation and version control config
README for additional insights
• Recommender pipeline:
– Data collection (TMDb CSVs) → feature engineering (“tags” column) → CountVectorizer → cosine similarity
– UI: Streamlit dropdown, recommendation display with posters
– Poster fetch: HTTP calls to TMDb API at runtime