Recommender system using pytorch geomtric to learn and predict an heterogeneous graph representation. Made for large networks: - Spark data management - distributed Deep graph network to predict new edges with labels (user rates a movie) - Sampling of the graph during epochs
Data source to unzip in data/01_raw : https://files.grouplens.org/datasets/movielens/ml-25m.zip
Use poetry install to install project's dependancies.
Run full pipe with kedro run
Run with kedro run --from_nodes training_start