基于分级注意力网络和多层对比学习的社交推荐

摘要/Abstract

参考文献

Metrics

doi:10.16088/j.issn.1001-6600.2024071801

摘要： 将社交关系融入推荐系统中,能有效提高推荐质量。然而现实世界中用户的交互数据是稀疏和复杂的,如何更好地利用社交信息是关键问题。现有社交推荐模型没有充分探索高阶好友的影响,而且忽略了用户间的关系强度和不同种类的关系对用户的影响,导致推荐性能不佳。为了解决上述问题,本文提出一个基于分级注意力网络和层次化对比学习的社交推荐模型。具体来说,首先,依据用户间不同关系构建用户级超图,扩大节点聚合的感知范围,加深模型深度。然后,设计多级注意力网络更好地捕捉用户交互数据之间的关系和重要性,其中,视图级自注意力机制捕获好友对用户的影响以及项目间的关联程度,通道级注意力自适应地调整不同种类的关系对用户的影响。同时,引入层次化对比学习对数据进行增强,包括视图间和跨视图的第一层对比学习和针对高阶关系的第二层对比学习,多维度捕获数据的细微差距和高层次的抽象特征。最后,将所提出的模型在4个公开基准数据集上进行评估,结果表明本文模型Precision、Recall、NDCG较其他最优基线模型分别提升7.61%、11.05%、10.69%,验证了本文模型的有效性。

关键词: 社交推荐, 注意力网络, 超图学习, 对比学习, 推荐系统

Abstract: Incorporating social relationships into recommender systems can effectively improve recommendation quality. However, real-life interactions among users are sparse and complex. The effective utilization of social information is a key issue. The impact of high-order friends is not fully explored by existing social recommendation models, and the strength of relationships between users and the impact of different kinds of relationships on users are ignored, leading to sub-optimal recommendation performance. To address these issues, Multi-level attention networks and Hierarchical Contrastive Learning for social recommendation (MHCL) is proposed. Specifically, the user-level hypergraph is first constructed based on different relationships between users to expand the perceptual scope of node aggregation and deepen the depth of the model. Then, a multilevel attention network is designed to better capture the relationships and importance between user interaction data, where the influence of friends on the user and the degree of inter-item relatedness are captured by the view-level self-attention mechanism, and the influence of different kinds of relationships on the user is adaptively adjusted by the channel-level attention. Meanwhile, hierarchical contrast learning is introduced to augment the data, including the first level of contrast learning between and across views and the second level of contrast learning for high-order relationships, to capture the subtle gaps and high-level abstract features of the data in multiple dimensions. Finally, the proposed model is evaluated on four publicly available benchmark datasets, and the effectiveness and reasonableness of MHCL are validated by the evaluation results. Social denoising will be the focus of future research to improve recommendation systems based on hypergraph neural networks.

Key words: social recommendation, attention networks, hypergraph learning, contrastive learning, recommender systems

中图分类号: TP391.1

张丽杰, 王绍卿, 张尧, 孙福振. 基于分级注意力网络和多层对比学习的社交推荐[J]. 广西师范大学学报（自然科学版）, 2025, 43(2): 133-148.

ZHANG Lijie, WANG Shaoqing, ZHANG Yao, SUN Fuzhen. Multi-level Attention Networks and Hierarchical Contrastive Learning for Social Recommendation[J]. Journal of Guangxi Normal University(Natural Science Edition), 2025, 43(2): 133-148.

[1] WU L, LI J W, SUN P J, et al. DiffNet++: a neural influence and interest diffusion network for social recommendation[J]. IEEE Transactions on Knowledge and Data Engineering, 2022, 34(10): 4753-4766. DOI: 10.1109/TKDE.2020.3048414.
[2] FAN W Q, MA Y, LI Q, et al. Graph neural networks for social recommendation[C] // Proceedings of the World Wide Web Conference WWW 2019. New Yerk, NY: Association for Computing Machinery, 2019: 417-426. DOI: 10.1145/3308558.3313488.
[3] WU L, SUN P J, FU Y J, et al. A neural influence diffusion model for social recommendation[C] // Proceedings of the 42nd International ACM SIGIR Conference on Research and Development in Information Retrieval. New York, NY: Association for Computing Machinery, 2019: 235-244. DOI: 10.1145/3331184.3331214.
[4] YU J L, YIN H Z, LI J D, et al. Self-supervised multi-channel hypergraph convolutional network for social recommendation[C] // Proceedings of the World Wide Web Conference WWW 2021. New York, NY: Association for Computing Machinery, 2021: 413-424. DOI: 10.1145/3442381.3449844.
[5] HAN J D, TAO Q, TANG Y F, et al. DH-HGCN: dual homogeneity hypergraph convolutional network for multiple social recommendations[C] // Proceedings of the 45th international ACM SIGIR conference on research and development in information retrieval. New York, NY: Association for Computing Machinery, 2022: 2190-2194. DOI: 10.1145/3477495.3531828.
[6] SUN Y D, ZHU D J, DU H W, et al. Motifs-based recommender system via hypergraph convolution and contrastive learning[J]. Neurocomputing, 2022, 512: 323-338. DOI: 10.1016/J.NEUCOM.2022.09.102.
[7] WANG T L, XIA L H, HUANG C. Denoised self-augmented learning for social recommendation[C] // Proceedings of the Thirty-Second International Joint Conference on Artificial Intelligence (IJCAI-23). Macao: IJCAI, 2023: 2324-2331. DOI: 10.24963/IJCAI.2023/258.
[8] YU J L, Yin H Z, Gao M, et al. Socially-aware self-supervised tri-training for recommendation[C] // Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining. New York, NY: Association for Computing Machinery, 2021: 2084-2092. DOI: 10.1145/3447548.3467340.
[9] WU Z H, PAN S R, CHEN F W, et al. A comprehensive survey on graph neural networks[J]. IEEE Transactions on Neural Networks and Learning Systems, 2021, 32(1): 4-24. DOI: 10.1109/TNNLS.2020.2978386.
[10] ZHANG Y Y, ZHU J W, ZHANG Y L, et al. Social-aware graph contrastive learning for recommender systems[J]. Applied Soft Computing, 2024, 158: 111558. DOI: 10.1016/j.asoc.2024.111558.
[11] JIANG W, GAO X Y, XU G D, et al. Challenging low homophily in social recommendation[C] // Proceedings of the ACM Web Conference 2024. New York, NY: Association for Computing Machinery, 2024: 3476-3484. DOI: 10.1145/3589334.3645460.
[12] 赵文涛, 刘甜甜, 薛赛丽, 等. 面向多视图融合的用户一致性社交推荐[J]. 计算机工程与应用, 2024, 60(10): 156-163. DOI: 10.3778/j.issn.1002-8331.2301-0099.
[13] 杨兴耀, 马帅, 张祖莲, 等. 基于偏好感知的去噪图卷积网络社交推荐[J]. 计算机工程, 2024, 50(10): 154-163. DOI: 10.19678/j.issn.1000-3428.0068748.
[14] 吴相帅, 孙福振, 张文龙, 等. 基于图注意力的异构图社交推荐网络[J]. 计算机应用研究, 2023, 40(10): 3076-3081, 3106. DOI: 10.19734/j.issn.1001-3695.2023.03.0085.
[15] 李邵莹, 孟丹, 孔超, 等. 面向社交推荐的自适应高阶隐式关系建模[J]. 软件学报, 2023, 34(10): 4851-4869. DOI: 10.13328/j.cnki.jos.006662.
[16] BU J J, TAN S L, CHEN C, et al. Music recommendation by unified hypergraph: combining social media information and music content[C] // Proceedings of the 18th ACM International Conference on Multimedia. New York, NY: Association for Computing Machinery, 2010: 391-400. DOI: 10.1145/1873951.1874005.
[17] WANG J L, DING K Z, HONG L J, et al. Next-item recommendation with sequential hypergraphs[C] // Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval. New York, NY: Association for Computing Machinery, 2020: 1101-1110. DOI: 10.1145/3397271.3401133.
[18] XIA X, YIN H Z, YU J L, et al. Self-supervised hypergraph convolutional networks for session-based recommendation[J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2021, 35(5): 4503-4511. DOI: 10.1609/aaai.v35i5.16578.
[19] 漆盛, 高榕, 邵雄凯, 等. 面向超图的可解释性对比元路径群组推荐[J]. 计算机工程与应用, 2024, 60(11): 268-280. DOI: 10.3778/j.issn.1002-8331.2304-0290.
[20] 傅晨波, 陈殊杭, 胡剑波, 等. 基于超图嵌入和有限注意力的社会化推荐[J]. 小型微型计算机系统, 2024, 45(1): 115-122. DOI: 10.20009/j.cnki.21-1106/TP.2022-0342.
[21] JIANG Y Q, HUANG C, HUANG L H. Adaptive graph contrastive learning for recommendation[C] // Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. New York, NY: Association for Computing Machinery, 2023: 4252-4261. DOI: 10.1145/3580305.3599768.
[22] XIA L H, HUANG C, HUANG C Z, et al. Automated self-supervised learning for recommendation[C] // Proceedings of the ACM Web Conference 2023. New York, NY: Association for Computing Machinery, 2023: 992-1002. DOI: 10.1145/3543507.3583336.
[23] MA G F, YANG X H, LONG H X, et al. Robust social recommendation based on contrastive learning and dual-stage graph neural network[J]. Neurocomputing, 2024, 584: 127597. DOI: 10.1016/J.NEUCOM.2024.127597.
[24] FENG Y F, YOU H X, ZHANG Z Z, et al. Hypergraph neural networks[J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2019, 33(1): 3558-3565. DOI: 10.1609/aaai.v33i01.33013558.
[25] ZHANG J W, GAO M, YU J L, et al. Double-scale self-supervised hypergraph learning for group recommendation[C] // Proceedings of the 30th ACM International Conference on Information & Knowledge Management. New York, NY: Association for Computing Machinery, 2021: 2557-2567. DOI: 10.1145/3459637.3482426.
[26] WEI T X, YOU Y N, CHEN T L, et al. Augmentations in hypergraph contrastive learning: Fabricated and generative[C] // Advances in Neural Information Processing Systems 35 (NeurIPS 2022). Red Hook, NY: Curran Associates Inc., 2022: 1909-1922.
[27] XIA L H, HUANG C, XU Y, et al. Hypergraph contrastive collaborative filtering[C] // Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval. New York, NY: Association for Computing Machinery, 2022: 70-79. DOI: 10.1145/3477495.353205.
[28] 王玉洁, 杨哲. 融合残差网络的自监督社交推荐算法[J]. 计算机科学与探索, 2024, 18(12): 3175-3188. DOI: 10.3778/j.issn.1673-9418.2401006.
[29] 王永贵, 陈书铭, 刘义海, 等. 结合超图对比学习和关系聚类的知识感知推荐算法[J]. 计算机科学与探索, 2024, 18(8): 2140-2155. DOI: 10.3778/j.issn.1673-9418.2305058.
[30] 胡海波, 杨丹, 聂铁铮, 等. 融入多影响力与偏好的图对比学习社交推荐算法[J]. 计算机科学, 2024, 51(7): 146-155. DOI: 10.11896/jsjkx.230400147.
[31] 张尧, 王绍卿, 吴瑕, 等. 面向捆绑推荐的解耦图对比学习[J/OL]. 计算机工程与应用[2024-07-18]. https://link.cnki.net/urlid/11.2127.TP.20240710.2137.007.
[32] MILO R, SHEN-ORR S, ITZKOVITZ S, et al. Network motifs: simple building blocks of complex networks[J]. Science, 2002, 298(5594): 824-827. DOI: 10.1515/9781400841356.217.
[33] GAO T Y, YAO X C, CHEN D Q. SimCSE: simple contrastive learning of sentence embeddings[C] // Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing. Stroudsburg,PA: Association for Computational Linguistics, 2021: 6894-6910. DOI: 10.18653/v1/2021.emnlp-main.552.
[34] DONG X P, SHEN J B. Triplet loss in Siamese network for object tracking[C] // Computer Vision-ECCV 2018: LNCS Volume 11217. Cham: Springer, 2018: 472-488. DOI: 10.1007/978-3-030-01261-8_28.
[35] VAN DEN OORD A, LI Y Z, VINYALS O. Representation learning with contrastive predictive coding[EB/OL]. (2019-01-22)[2024-07-18]. https://arxiv.org/abs/1807.03748. DOI: 10.48550/arXiv.1807.03748.
[36] RENDLE S, FREUDENTHALER C, GANTNER Z, et al. BPR: Bayesian personalized ranking from implicit feedback[C] // Proceedings of the 25th Conference on Uncertainty in Artificial Intelligence. Arlington, VA: AUAI Press, 2009: 452-461.
[37] ZHAO T, MCAULEY J, KING I. Leveraging social connections to improve personalized ranking for collaborative filtering[C] // Proceedings of the 23rd ACM International Conference on Information and Knowledge Management. New York, NY: Association for Computing Machinery, 2014: 261-270. DOI: 10.1145/2661829.2661998.
[38] HE X N, DENG K, WANG X, et al. LightGCN: simplifying and powering graph convolution network for recommendation[C] // Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval. New York, NY: Association for Computing Machinery, 2020: 639-648. DOI: 10.1145/3397271.3401063.

Just accepted

Online first

Just accepted

Online first

Viewed

Full text

Abstract

Just accepted	Online first	Issue

0	0	4

From	Others	local

Times	3	1
Rate	75%	25%

Cited

Web of Science	Crossref	ScienceDirect	Search for Citations in Google Scholar >>


This page requires you have already subscribed to WoS.

Shared

Discussed