Graph Heterogeneous Multi-Relational Recommendation

Traditional studies on recommender systems usually leverage only one type of user behaviors (the optimization target, such as purchase), despite the fact that users also generate a large number various types interaction data (e.g., view, click, add-to-cart, etc). Generally, these heterogeneous multi-relational provide well-structured information and can be used for high-quality recommendation. Early efforts towards leveraging fail to capture high-hop structure user-item interactions, which are unable make full use them may achieve constrained recommendation performance. In this work, we propose new model named Graph Heterogeneous Collaborative Filtering (GHCF). To explore take advantages Convolutional Network (GCN) further improve it jointly embed both representations nodes (users items) relations prediction. Moreover, fully utilize whole data, perform advanced efficient non-sampling under multi-task learning framework. Experimental results two public benchmarks show GHCF significantly outperforms state-of-the-art methods, especially cold-start who have few primary item interactions. Further analysis verifies importance proposed embedding propagation modelling showing rationality effectiveness GHCF. Our implementation has been released (https://github.com/chenchongthu/GHCF).