Content-Based Filtering Demo
Content-based filtering recommends items similar to what you've liked in the past. The algorithm analyzes the attributes of movies you rate highly (genre, director, mood, etc.) and finds other movies with similar characteristics.
Why is this transparent? Unlike collaborative filtering ("users like you also liked..."), content-based recommendations are fully explainable. You can see exactly which attributes match your preferences!
The Algorithm: For each unrated movie, we calculate a similarity score based on:
• Genre match (30%): Do you like Action? We'll recommend more Action!
• Director match (20%): Loved a Nolan film? Here's another!
• Mood match (20%): Prefer dark films? We'll find similar tones.
• Length match (10%): Do you prefer shorter or longer films?
• Era match (10%): Classic or modern? We track your preferences.
• Critic ratings (10%): We weight by quality to recommend well-reviewed films.
⚠️ The Filter Bubble: Notice how recommendations stay similar to what you've rated? This is the "filter bubble" effect - content-based systems can limit discovery. Try rating movies from different genres to see how the algorithm adapts!