Overview:

The Letterboxd Recommendations project scrapes Letterboxd data to create a movie recommendation model based on user ratings. This model uses collaborative filtering to suggest movies without considering genres or themes. The project includes a popularity filter to offer obscure recommendations and can return 50 suggestions per query.

Features:

Scrapes Letterboxd data: Collects user ratings from Letterboxd profiles.
Collaborative filtering model: Uses user ratings to suggest movies based on similarities in rating patterns.
Popularity filter: Filters recommendations based on the number of ratings a movie has received.
Returns 50 recommendations: Provides a list of 50 movie suggestions per query.

Summary:

The Letterboxd Recommendations project offers a movie recommendation model based on user ratings scraped from Letterboxd profiles. It utilizes collaborative filtering to provide suggestions without considering genres or themes, including a popularity filter to offer obscure recommendations. The project limits the sample size to 500,000 samples due to time and computing constraints but can still return 50 recommendations, though occasionally with some misses. To run the project, users need to set up a local MongoDB server and follow the provided instructions for crawling, training, and model execution.

Sdl60660

Letterboxd_recommendations

River Runner