TMHGCN：一种时序多通道超图卷积电商推荐方法
TMHGCN: A Time Series Multi-Channel Hypergraph Convolution E-Commerce Recommendation Method
摘要: 随着信息化时代的发展，人们的交互强度与日俱增，在电商推荐系统中，传统推荐方法已经无法适应当今现实环境。人们通过处理社交关系来增强推荐质量，大量研究表明，社交关系的引入可以在一定程度上缓解冷启动和稀疏性问题，但真实场景中的用户关联并非简单的“一对一”，还存在着社区关系、共同兴趣圈等复杂社交关系，同时用户的兴趣还可能会随着时间发生改变。超图卷积对于刻画这种“多对多”的复杂社交关系效率较高，但如何在处理复杂关系的同时可以减小噪声影响，同时又不忽视用户兴趣的动态变化成为了社交系统中一个值得进一步探索的方向。基于此，本文提出了一种时序多通道超图卷积网络(TMGCHN)：在技术层面，我们通过将不同类型的关系数据划分为多个通道(比如社交、交互或二者结合等)，并在每个通道上构建专注于该类型数据的超图；同时引入时序编码模块对用户特征进行处理，学习用户兴趣的动态变化。通过多通道超图的信息聚合，TMGHCN可以更全面、更细致地预测用户偏好，而在Yelp、Douban等大量真实数据集上的实验结果表明，该模型在多项指标上均取得稳步提升。
Abstract: With the development of information age, people’s interaction intensity is increasing day by day. In the e-commerce recommendation system, the traditional recommendation method has been unable to adapt to today’s realistic environment. People enhance the recommendation quality by dealing with social relationships. A large number of studies show that the introduction of social relationships can alleviate the cold start and sparseness problems to some extent, but the user association in real scenes is not simple “one-on-one”, there are also complex social relationships, such as community relationships and common interest circles, and users’ interests may change over time. Hypergraph Convolutional Networks have a lot of experience in depicting this “many-to-many” complex social relationship, but how to deal with the complex relationship while reducing the noise impact without ignoring the dynamic changes of users’ interests has become a direction worthy of further exploration in social systems. Based on this, this paper proposes a time series multi-channel hypergraph convolution network (TMGCHN): On the technical level, we divide different types of relational data into multiple channels (such as social, interactive or a combination of the two), and build a hypergraph focusing on this type of data on each channel; at the same time, the sequential coding module is introduced to process user characteristics and learn the dynamic changes of user interests. Through the information aggregation of multi-channel hypergraph, TMGHCN can predict user preferences more comprehensively and carefully, and the experimental results on a large number of real data sets, such as Yelp and Douban, show that the model has made steady progress in many indicators.
文章引用：程实, 邵逸飞, 张志鑫, 顾卫江. TMHGCN：一种时序多通道超图卷积电商推荐方法[J]. 电子商务评论, 2026, 15(5): 915-924. https://doi.org/10.12677/ecl.2026.155593

参考文献

[1] Feng, Y., You, H., Zhang, Z., Ji, R. and Gao, Y. (2019) Hypergraph Neural Networks. Proceedings of the AAAI Conference on Artificial Intelligence, 33, 3558-3565. [Google Scholar] [CrossRef] 
[2] Bai, S., Zhang, F. and Torr, P.H.S. (2021) Hypergraph Convolution and Hypergraph Attention. Pattern Recognition, 110, Article 107637. [Google Scholar] [CrossRef] 
[3] Bu, J., Tan, S., Chen, C., Wang, C., Wu, H., Zhang, L., et al. (2010) Music Recommendation by Unified Hypergraph: Combining Social Media Information and Music Content. Proceedings of the 18th ACM international conference on Multimedia, Firenze, 25-29 October, 391-400. [Google Scholar] [CrossRef] 
[4] 成耀, 张铎, 周宇, 何金凤, 程实. 基于模糊聚类的电商企业不平衡财务数据风险预测方法[J]. 电子商务评论, 2025, 14(1): 640-647.
[5] Ji, S., Feng, Y., Ji, R., Zhao, X., Tang, W. and Gao, Y. (2020) Dual Channel Hypergraph Collaborative Filtering. Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, Online, 6-10 July 2020, 2020-2029. [Google Scholar] [CrossRef] 
[6] Liu, X., Meng, S., Li, Q., Xu, X., Qi, L., Dou, W., et al. (2023) Disentangled Hypergraph Collaborative Filtering for Social Recommendation. 2023 IEEE International Conference on Web Services (ICWS), Chicago, 2-8 July 2023, 475-482. [Google Scholar] [CrossRef] 
[7] Yang, D., Qu, B., Yang, J. and Cudre-Mauroux, P. (2020) LBSN2vec++: Heterogeneous Hypergraph Embedding for Location-Based Social Networks. IEEE Transactions on Knowledge and Data Engineering, 34, 1843-1855. [Google Scholar] [CrossRef] 
[8] He, X., Deng, K., Wang, X., Li, Y., et al. (2020) LightGCN: Simplifying and Powering Graph Convolution Network for Recommendation. Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval, New York, 25-30 July 2020, 639-648.
[9] Rendle, S., Freudenthaler, C., Gantner, Z. and Schmidt-Thieme, L. (2009) BPR: Bayesian Personalized Ranking from Implicit Feedback. Proceedings of the 25th Conference on Uncertainty in Artificial Intelligence, Arlington, 18-21 June 2009, 452-461.
[10] Berg, R.V.D., Kipf, T.N. and Welling, M. (2017) Graph Convolutional Matrix Completion. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, New York, 13-17 August 2017, 1-10.
[11] Xia, X., Yin, H., Yu, J., Wang, Q., Cui, L. and Zhang, X. (2021) Self-Supervised Hypergraph Convolutional Networks for Session-Based Recommendation. Proceedings of the AAAI Conference on Artificial Intelligence, 35, 4503-4511. [Google Scholar] [CrossRef] 
[12] Fan, W., Ma, Y., Li, Q., He, Y., Zhao, E., Tang, J., et al. (2019) Graph Neural Networks for Social Recommendation. The World Wide Web Conference, San Francisco, 13-17 May, 417-426. [Google Scholar] [CrossRef] 

为你推荐