Self-Supervised Multi-Channel Hypergraph Convolutional Network for Social Recommendation

By Junliang Yu, Hongzhi Yin, Jundong Li, Qinyong Wang, Nguyen Quoc Viet Hung, Xiangliang Zhang. The Web Conference, 2021

617 citations recorded by the source corpus.

Social relations are often used to improve recommendation quality when user-item interaction data is sparse in recommender systems. Most existing social recommendation models exploit pairwise relations to mine potential user preferences. However, real-life interactions among users are very complex and user relations can be high-order. Hypergraph provides a natural way to model high-order relations, while its potentials for improving social recommendation are under-explored. In this paper, we fill this gap and propose a multi-channel hypergraph convolutional network to enhance social recommendation by leveraging high-order user relations. Technically, each channel in the network encodes a hypergraph that depicts a common high-order user relation pattern via hypergraph convolution. By aggregating the embeddings learned through multiple channels, we obtain comprehensive user representations to generate recommendation results. However, the aggregation operation might also obscure the inherent characteristics of different types of high-order connectivity information. To compensate for the aggregating loss, we innovatively integrate self-supervised learning into the training of the hypergraph convolutional network to regain the connectivity information with hierarchical mutual information maximization. Extensive experiments on multiple real-world datasets demonstrate the superiority of the proposed model over the current SOTA methods, and the ablation study verifies the effectiveness and rationale of the multi-channel setting and the self-supervised task. The implementation of our model is available via https://github.com/Coder-Yu/RecQ.

Subfields: Artificial Intelligence, Information Systems, Social Psychology

View Paper: https://www.semanticscholar.org/paper/68c9d777b66c24967a7c808210093684d89eba10

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Hongzhi Yin
2. Tong Chen
3. Xin Xia
4. Junliang Yu
5. Xiangnan He
6. Chao Huang
7. Lianghao Xia
8. Yong Li

In Information Systems

1. Hongzhi Yin
2. Tong Chen
3. Xin Xia
4. Junliang Yu
5. Chao Huang
6. Xiangnan He
7. Lianghao Xia
8. Yong Li

In Social Psychology

1. Hongzhi Yin
2. Srijan Kumar

Citations in the corpus

• Hypergraph Contrastive Collaborative Filtering (2022, methodology): “• MHCN [45].”
• A Survey of Graph Neural Networks for Recommender Systems: Challenges, Methods, and Directions (2021, methodology): “Moreover, some studies, such as MHCN [227] and HOSR [113], attempt to enhance the recommendation by modeling the high-order relations in social networks more suiciently.”
• Contrastive Learning for Representation Degeneration Problem in Sequential Recommendation (2021, methodology): “DHCN [50] and MHCN [56] are graph-based methods with contrastive learning on node-level representation.”
• Graph Neural Networks in Recommender Systems: A Survey (2020, background): “NDCG DiffNet [171], DiffNet++ [170], ESRF [190], DGRec [129], GNN-SoR [46], KCGN [61], MHCN [191]”
• Self-Supervised Learning for Recommender Systems: A Survey (2022, methodology): “Compared to MHCN, HCCF [95] proposes to parameterize hypergraph dependency matrices instead of manually defining hypergraph structures.”
• XSimGCL: Towards Extremely Simple Graph Contrastive Learning for Recommendation (2022, background): “Inspired by the success of CL in other fields, the community also has started to integrate CL into recommendation [12], [13], [14], [15], [16], [17], [59].”
• LightGCL: Simple Yet Effective Graph Contrastive Learning for Recommendation (2023, background): “They also perform better than some of other self-supervised learning approaches (MHCN).”
• Are Graph Augmentations Necessary?: Simple Graph Contrastive Learning for Recommendation (2021, methodology): “Because we focus on the Top-N recommendation, following the convention in the previous research [41, 41], we discard ratings less than 4 in DoubanBook, which is with a 1-5 rating scale, and reset the rest to 1.”
• Knowledge Graph Contrastive Learning for Recommendation (2022, methodology): “To investigate the robustness of our KGCL in handing users without sufficient interactions, we follow similar settings in [50] to generate sparse user set with less than 20 interactions for Yelp2018 and Amazon-Book, and 5 interactions for MIND data.”
• Graph Self-Supervised Learning: A Survey (2021, background): “S(2)-MHCN [115] and DHCN [116] employ contrastive tasks for hypergraph representation learning for social- and session- based recommendation, respectively.”
• A Survey of Graph Neural Networks for Social Recommender Systems (2022, methodology): “…MutualRec [110], ATGCN [74], DiffNet [105], DiffNet++ [104], DiffNetLG [80], MEGCN [33], GLOW [36], SCGRec [117], SEPT [120], DcRec [103], MHCN [122], DH-HGCN [24], GSFR [109], HOSR [50] IGRec [10], ESRF [121], SoHRML [51], DSR [75], SRAN [112], IDiffNet [41], CGL [127], MrAPR [81], DISGCN…”