A content-based movie recommender combining cosine similarity, TMDB metadata, IMDb review scraping, and NLP sentiment analysis in a Flask application.
Impact
content similarity plus review sentiment in one movie discovery experience
Hiring signal
Classical ML fundamentals, feature engineering, similarity search, external API integration, web scraping, NLP inference, and Flask delivery.
My role
Developer · data preparation, recommendation logic, Flask backend, TMDB integration, IMDb review scraping, sentiment model integration, and UI flow
Status
Source or architecture details are available where applicable.
The problem
Movie discovery is often limited to generic popularity rankings. The objective was to recommend titles that are content-similar to a selected movie while also presenting metadata, cast information, and review sentiment.
My approach
The engine constructs a CountVectorizer feature matrix from combined movie attributes and applies cosine similarity to rank the ten nearest titles. A Flask backend serves recommendations, TMDB supplies movie and cast metadata, and BeautifulSoup extracts IMDb reviews for classification through persisted NLP vectorizer and model artifacts.
Architecture judgment
Content-based filtering operates without user-interaction history and provides an interpretable similarity mechanism: titles are ranked according to the angular proximity of their engineered feature vectors.
Architecture
The complete execution path, service boundaries, data stores, security controls, validation stages, and operational telemetry.
Implementation
Technology
Interactive trace
Run the trace to see the main workflow stages.