GNN-Based Collaborative Filtering Approach for Movie Recommender System

Movie recommendation systems enhance user experience by providing personalized content. This paper introduces a novel graph-based collaborative filtering approach, leveraging graph models to capture complex user-item relationships. Our method, tested on the MovieLens 100K and 1M datasets, improves predictive accuracy, achieving a 14.8% improvement in MAE and 1.0% in RMSE on 100K. For MovieLens-1M, it achieves an MAE of 0.658 and RMSE of 0.887, outperforming other methods. The method achieves the best results on 100K with the lowest MAE (0.618) and RMSE (0.918), demonstrating its effectiveness in providing accurate and robust movie recommendations.