amithjoel95823/GNN-for-Recommendation-System ? reverse-engineered prompt

Build me a movie recommendation app that uses graph neural networks along with collaborative filtering and content based filtering.

I want it to train on the MovieLens 25M data, learn from user ratings and movie details like genres, then predict how much a user will like a movie and recommend good matches. Please include a clear training script and a test script, plus a notebook I can use to explore the data and results. Make it easy to set up from a requirements file, and make sure the code is organized in a simple way with reusable pieces for the dataset, model, training, testing, and helper functions.

Also add basic evaluation so I can see metrics like RMSE, precision, recall, and F1 score, and show a couple of simple plots for true ratings and prediction errors. If anything is unclear, check current docs online and keep the implementation practical and easy to run.