基于邻域采样的多任务图推荐算法

doi:10.3778/j.issn.1002-8331.2308-0212

摘要/Abstract

摘要： 近年来，图神经网络（GNN）成为解决协同过滤的主流方法之一。它通过构建用户-物品图，模拟用户与物品的交互关系，并用GNN学习它们的特征表示。尽管现有在模型结构上的研究已取得了较大进展，但如何在图结构上更有效地进行负采样仍未有效解决。为此，提出一种基于邻域采样的多任务图推荐算法。该算法提出了一种基于GNN的邻域采样策略，该策略以每个用户为中心构建子图，将次高阶物品作为用户邻域采样的负样本，可以更有效地挖掘强负样本并提高采样质量。通过GNN对图结点进行信息聚合与特征提取，得到结点的最终嵌入表示。设计一种余弦边际损失来过滤部分冗余负样本，以有效减少采样过程中的噪声数据。同时，该算法引入了多任务策略对模型进行联合优化，以增强模型的泛化能力。在3个公开数据集上进行的大量实验表明，该算法在大多数情况下明显优于其他主流算法。

关键词: 图神经网络, 协同过滤, 负采样, 邻域采样, 余弦边际损失, 多任务策略

Abstract: In recent years, graph neural network (GNN) has become a mainstream method for collaborative filtering. It constructs user-item graphs to simulate interactions and utilizes GNN to learn their features. Although there have been significant advancements in model structures, effective negative sampling on graph structures remains challenging. To address this issue, a multi-task graph recommendation algorithm based on neighborhood sampling is proposed. Firstly, the algorithm introduces a neighborhood sampling strategy based on GNN, centering each subgraph around individual users and using higher-order items as negative samples for user neighborhood sampling. This approach effectively explores strong negative instances and enhances sampling quality. Secondly, GNN is employed to aggregate information and extract features from graph nodes, obtaining the final node embeddings. Finally, a margin loss of cosine is designed to filter redundant negative samples, effectively reducing noise in the sampling process. Additionally, the algorithm incorporates a multi-task strategy to jointly optimize the model, enhancing its generalization ability. Extensive experiments conducted on three public datasets demonstrate that this algorithm outperforms other mainstream methods in the majority of cases.

Key words: graph neural network (GNN), collaborative filtering, negative sampling, neighborhood sampling, margin loss of cosine, multi-task strategy

张俊三, 肖森, 高慧, 邵明文, 张培颖, 朱杰. 基于邻域采样的多任务图推荐算法[J]. 计算机工程与应用, 2024, 60(9): 172-180.

ZHANG Junsan, XIAO Sen, GAO Hui, SHAO Mingwen, ZHANG Peiying, ZHU Jie. Multi-Task Graph Recommendation Algorithm Based on Neighborhood Sampling[J]. Computer Engineering and Applications, 2024, 60(9): 172-180.

参考文献

[1] HUANG J T, SHARMA A, SUN S, et al. Embedding-based retrieval in facebook search[C]//Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, 2020: 2553-2561.
[2] KOREN Y, BELL R, VOLINSKY C. Matrix factorization techniques for recommender systems[J]. Computer, 2009, 42(8): 30-37.
[3] MA H, YANG H, LYU M R, et al. Sorec: social recommendation using probabilistic matrix factorization[C]//Proceedings of the 17th ACM Conference on Information and Knowledge Management, 2008: 931-940.
[4] HE X, LIAO L, ZHANG H, et al. Neural collaborative filtering[C]//Proceedings of the 26th International Conference on World Wide Web, 2017: 173-182.
[5] BERG R, KIPF T N, WELLING M. Graph convolutional matrix completion[J]. arXiv:1706.02263, 2017.
[6] WANG X, HE X, WANG M, et al. Neural graph collaborative filtering[C]//Proceedings of the 42nd International ACM SIGIR Conference on Research and Development in Information Retrieval, 2019: 165-174.
[7] HE X, 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, 2020: 639-648.
[8] CASELLES-DUPRé H, LESAINT F, ROYO-LETELIER J. Word2vec applied to recommendation: hyperparameters matter[C]//Proceedings of the 12th ACM Conference on Recommender Systems, 2018: 352-356.
[9] CAI T T, FRANKLE J, SCHWAB D J, et al. Are all negatives created equal in contrastive instance discrimination?[J]. arXiv:2010.06682, 2020.
[10] WU Z, PAN S, CHEN F, et al. A comprehensive survey on graph neural networks[J]. IEEE Transactions on Neural Networks and Learning Systems, 2020, 32(1): 4-24.
[11] PAN L M, XIE Y X, FENG Y S, et al. Semantic graphs for generating deep questions[J]. arXiv:2004.12704, 2020.
[12] GAO C, ZHENG Y, LI N, et al. A survey of graph neural networks for recommender systems: challenges, methods, and directions[J]. ACM Transactions on Recommender Systems, 2023, 1(1): 1-51.
[13] 刘鑫, 梅红岩, 王嘉豪, 等. 图神经网络推荐方法研究[J]. 计算机工程与应用, 2022, 58(10): 41-49.
LIU X, MEI H Y, WANG J H, et al. Research on graph neural network recommendation method[J]. Computer Engineering and Applications, 2022, 58(10): 41-49.
[14] MAO K, ZHU J, XIAO X, et al. UltraGCN: ultra simplification of graph convolutional networks for recommendation[C]//Proceedings of the 30th ACM International Conference on Information & Knowledge Management, 2021: 1253-1262.
[15] LIU F, CHENG Z, ZHU L, et al. Interest-aware message-passing GCN for recommendation[C]//Proceedings of the Web Conference, 2021: 1296-1305.
[16] HUANG C, XU H, XU Y, et al. Knowledge-aware coupled graph neural network for social recommendation[C]//Proceedings of the AAAI Conference on Artificial Intelligence, 2021: 4115-4122.
[17] RENDLE S, FREUDENTHALER C, GANTNER Z, et al. BPR: Bayesian personalized ranking from implicit feedback[J]. arXiv:1205.2618, 2012.
[18] MAO K, ZHU J, WANG J, et al. SimpleX: a simple and strong baseline for collaborative filtering[C]//Proceedings of the 30th ACM International Conference on Information & Knowledge Management, 2021: 1243-1252.
[19] YANG Z, DING M, ZOU X, et al. Region or global a principle for negative sampling in graph-based recommendation[J]. IEEE Transactions on Knowledge and Data Engineering, 2023, 35(6): 6264-6277.
[20] DING J, QUAN Y, YAO Q, et al. Simplify and robustify negative sampling for implicit collaborative filtering[C]//Proceedings of the 34th International Conference on Neural Information Processing Systems, 2020: 1094-1105.
[21] HUANG T, DONG Y, DING M, et al. MixGCF: an improved training method for graph neural network-based recommender systems[C]//Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining, 2021: 665-674.
[22] LINDEN G, SMITH B, YORK J. Amazon. com recommendations: item-to-item collaborative filtering[J]. IEEE Internet Computing, 2003, 7(1): 76-80.
[23] KIPF T N, WELLING M. Semi-supervised classification with graph convolutional networks[J]. arXiv:1609.02907, 2016.
[24] XU K, LI C, TIAN Y, et al. Representation learning on graphs with jumping knowledge networks[C]//International Conference on Machine Learning, 2018: 5453-5462.
[25] ANAGNOSTOPOULOS A, KUMAR R, MAHDIAN M. Influence and correlation in social networks[C]//Proceedings of the 14th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 2008: 7-15.
[26] SHARMA K, LEE Y C, NAMBI S, et al. A survey of graph neural networks for social recommender systems[J]. arXiv:2212.04481, 2022.
[27] HE R, MCAULEY J. Ups and downs: modeling the visual evolution of fashion trends with one-class collaborative filtering[C]//Proceedings of the 25th International Conference on World Wide Web, 2016: 507-517.
[28] HE R, MCAULEY J. VBPR: visual Bayesian personalized ranking from implicit feedback[C]//Proceedings of the AAAI Conference on Artificial Intelligence, 2016.
[29] WANG X, JIN H, ZHANG A, et al. Disentangled graph collaborative filtering[C]//Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval, 2020: 1001-1010.
[30] BUDHIRAJA A. Revisiting SVD to generate powerful node embeddings for recommendation systems[J]. arXiv:2110.
03665, 2021.
[31] FAN W, LIU X, JIN W, et al. Graph trend filtering networks for recommendation[C]//Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval, 2022: 112-121.
[32] KINGMA D P, BA J. Adam: a method for stochastic optimization[J]. arXiv:1412.6980, 2014.