Self-supervised Graph Learning for Recommendation

By Jiancan Wu, Xiang Wang, Fuli Feng, Xiangnan He, Liang Chen, Jianxun Lian, Xing Xie. Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, 2020

1,757 citations recorded by the source corpus.

Representation learning on user-item graph for recommendation has evolved from using single ID or interaction history to exploiting higher-order neighbors. This leads to the success of graph convolution networks (GCNs) for recommendation such as PinSage and LightGCN. Despite effectiveness, we argue that they suffer from two limitations: (1) high-degree nodes exert larger impact on the representation learning, deteriorating the recommendations of low-degree (long-tail) items; and (2) representations are vulnerable to noisy interactions, as the neighborhood aggregation scheme further enlarges the impact of observed edges. In this work, we explore self-supervised learning on user-item graph, so as to improve the accuracy and robustness of GCNs for recommendation. The idea is to supplement the classical supervised task of recommendation with an auxiliary self-supervised task, which reinforces node representation learning via self-discrimination. Specifically, we generate multiple views of a node, maximizing the agreement between different views of the same node compared to that of other nodes. We devise three operators to generate the views --- node dropout, edge dropout, and random walk --- that change the graph structure in different manners. We term this new learning paradigm asSelf-supervised Graph Learning (SGL), implementing it on the state-of-the-art model LightGCN. Through theoretical analyses, we find that SGL has the ability of automatically mining hard negatives. Empirical studies on three benchmark datasets demonstrate the effectiveness of SGL, which improves the recommendation accuracy, especially on long-tail items, and the robustness against interaction noises. Our implementations are available at \urlhttps://github.com/wujcan/SGL.

Subfields: Artificial Intelligence, Information Systems

View Paper: https://www.semanticscholar.org/paper/75739ed2ddebd7982042f516f407553f8d3110f8

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. Xiangnan He
2. Hongzhi Yin
3. Wayne Xin Zhao
4. Tong Chen
5. Junliang Yu
6. Xiang Wang
7. Xin Xia
8. Yupeng Hou

In Information Systems

1. Hongzhi Yin
2. Xiangnan He
3. Tong Chen
4. Junliang Yu
5. Xin Xia
6. Xiang Wang
7. Wayne Xin Zhao
8. Chao Huang

Citations in the corpus

• A Survey of Graph Neural Networks for Recommender Systems: Challenges, Methods, and Directions (2021, methodology): “For example, Yu et al. [227] and Wu et al. [194] attempt to enhance GNN-based recommendation by designing auxiliary tasks from the graph structure with self-supervision.”
• Graph Neural Networks in Recommender Systems: A Survey (2020, background): “Inspired by the success of SSL in other areas, recent efforts have leveraged SSL to recommender systems, and made remarkable achievements [169, 208].”
• Improving Graph Collaborative Filtering with Neighborhood-enriched Contrastive Learning (2022, methodology): “For example, SGL [33] devise random data argumentation operator and construct the contrastive objective to improve the accuracy and robustness of GCNs for recommendation.”
• Representation Learning with Large Language Models for Recommendation (2023, methodology): “Specifically, we utilize LightGCN [11] as the underlying backbone recommender model, which generate a ranking list of 50 candidate items preferred by each user based on learned user-item interaction probabilities.”
• Hypergraph Contrastive Collaborative Filtering (2022, background): “Empirical results show that HCCF consistently improves the performance over state-of-the-art recommendation models, e . g ., average relative improvement of 24.7% over LightGCN, 33.8% over DHCF, 20.2% over MHCN and 14.8% over SGL in terms of NDCG@20.”
• Contrastive Learning for Representation Degeneration Problem in Sequential Recommendation (2021, methodology): “For the collaborative filtering methods, SGL [48] applies the NCE in node-level representation learning.”
• LightGCL: Simple Yet Effective Graph Contrastive Learning for Recommendation (2023, methodology): “To construct perturbed views, SGL (Wu et al., 2021) has been proposed to generate node pairs of positive view by corrupting the structural information of user-item interaction graph using stochastic augmentation strategies, e.”
• LLMRec: Large Language Models with Graph Augmentation for Recommendation (2023, background): “Future directions for investigation include integrating causal inference into side information debiasing and exploring counterfactual factors for context-aware user preference.”
• Bootstrap Latent Representations for Multi-modal Recommendation (2022, background): “1) Contrastive View Generator: Prior studies [16], [24], [37] use graph augmentations to generate two alternate views of the original graph for self-supervised learning.”
• UltraGCN: Ultra Simplification of Graph Convolutional Networks for Recommendation (2021, background): “Extensive experimental results demonstrate that UltraGCN achieves impressive improvements over the state-of-the-art CF models in terms of both accuracy and efficiency.”
• Self-Supervised Learning for Recommender Systems: A Survey (2022, methodology): “In SELFRec, we have incorporated several high-quality datasets that are widely used in the surveyed papers, such as Amazon-Book[31],Yelp-2018[84],andAmazon-Beauty[107], forbothgeneralandsequentialscenarios.”
• Graph Neural Networks for Recommender System (2022, methodology): “Moreover, the self-supervised paradigm is also useful in GNN-based models [22].”
• Knowledge Graph Contrastive Learning for Recommendation (2022, methodology): “Different from most existing contrastive learning models ( e.g., SGL [44], GraphCL [49]) which directly performs transformation on structure-level augmentations ( e.g., node dropping out or edge edge perturbation), we incorporate item knowledge semantics into a co-contrastive learning architecture.”
• Are Graph Augmentations Necessary?: Simple Graph Contrastive Learning for Recommendation (2021, methodology): “In this paper, we revisit the most commonly used dropoutbased augmentation on graphs [29, 36] which assumes that the representations are invariant to partial structure perturbations.”
• Intent Contrastive Learning for Sequential Recommendation (2022, methodology): “Contrastive Self-Supervised Learning (SSL) has brought much attentions by different research communities including CV [2, 4, 9, 14, 20] and NLP [7, 8, 29, 50], as well as recommendation [44, 45, 47, 51].”