Movie Recommender System

This repository contains the implementation of an advanced movie recommender system using LightGCN in PyTorch Geometric (PyG). The system integrates message passing and embeddings from a bipartite graph of users and movies, utilizing both supervised and self-supervised learning techniques to enhance recommendation quality.

Features

• Graph Neural Networks: Built using LightGCN in PyG, integrating message passing and embeddings.
• Learning Methods: Utilizes both supervised and self-supervised learning to improve recommendation quality.
• Loss Functions: Trained using BPR (Bayesian Personalized Ranking) and RMSE (Root Mean Squared Error) loss functions.
• Optimizer: Uses Adam optimizer with learning rate decay for optimized convergence.
• Data Processing: Efficiently processes and evaluates graph data with sparse matrices and edge indices.
• Evaluation Metrics: Uses recall, precision, and NDCG (Normalized Discounted Cumulative Gain) metrics for evaluation.

Installation

1. Clone the repository:

   git clone https://github.com/ravindramohith/movie_recommender_system.git

2. Navigate to the project directory:

   cd movie_recommender_system

Usage

• Supervised Learning:

  • The supervised learning implementation can be found in recommender-system-using-supervised-gnn_modified.ipynb.
  • Follow the notebook to train and evaluate the supervised recommender system.

• Self-Supervised Learning:

  • The self-supervised learning implementation can be found in recommender-system-using-self-supervised-gnn_modified.ipynb.
  • Follow the notebook to train and evaluate the self-supervised recommender system.

Evaluation

The model performance is evaluated using the following metrics:

• Recall: Measures the ability of the recommender system to capture relevant items.
• Precision: Measures the accuracy of the recommended items.
• NDCG: Measures the ranking quality of the recommendations.

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

Contact

For any questions or suggestions, please contact me through [email protected].