Recommendation System with Graph Neural Networks

A personalized movie recommendation system built with Graph Neural Networks, treating users and items as nodes in an interaction graph and learning from the structure of who-likes-what.

Approach

• Represents users, items, and their ratings as a graph, then learns node embeddings with a GNN to predict preferences.
• Generates personalized recommendations from the learned representations.
• A hands-on exploration of how graph-based learning compares to classic collaborative filtering for recommendation.

Code on GitHub.