Collaborative Filtering with Graph-Based Implicit Feedback

Introducing consumed items as users' implicit feedback in matrix factorization (MF) method, SVD++ is one of the most effective collaborative filtering methods for personalized recommender systems. However, powerful SVD++ has two limitations: (i) only the user-side implicit feedback is utilized instead of item-side implicit feedback which can also enrich item representations; (ii) in SVD++, the interacted items are equally weighted when combining the implicit feedback, which cannot reflect user's true preferences accurately. To tackle the above limitations, this paper proposes Graph-based collaborative filtering (GCF) model, Weighted Graph-based collaborative filtering (W-GCF) model and Attentive Graph-based collaborative filtering (A-GCF) model, which (i). generalize the implicit feedback to item side based on the user-item bipartite graph; (ii). flexibly learn the weights of individuals in the implicit feedback to improve the model's capacity. Comprehensive experiments show that the proposed models outperform state-of-the-art models. For sparse implicit feedback scenarios, additional improvement is further achieved by leveraging the two-step implicit feedback information.