双通道图协同过滤推荐算法

doi:10.19678/j.issn.1000-3428.0062453

摘要/Abstract

摘要： 协同过滤是一种应用广泛的推荐算法，其核心过程是学习用户和商品的向量表示。基于图卷积网络（GCN）的协同过滤算法在向量嵌入过程中加入邻居节点的关联信息，进一步提升了算法的推荐性能。然而，图协同过滤算法中存在过平滑现象，且其仅采用邻接矩阵在局部结构中扩展，没有从图的整体结构出发挖掘节点间潜在的交互模式，使得交互信息来源单一。提出一种基于GCN的双通道协同过滤推荐算法DCCF。将向量嵌入过程划分为局部卷积通道和全局卷积通道，以获取不同类型的连接信息。在局部卷积通道中，直接定位邻域节点并使用单层网络结构完成计算，优化信息的聚合方式以应对过平滑问题。在全局卷积通道中，通过聚类的方式构造全局交互图并参与信息的聚合过程，从而挖掘节点间的潜在联系。将局部信息与全局信息相结合，以获得包含不同类型高阶关系的节点向量表示。在3个公开数据集上进行对比实验，结果表明，相较基准算法中性能表现最优的模型，DCCF在归一化折损累计增益和召回率这2个指标上最高分别提升2.8%和5.0%。

关键词: 推荐算法, 深度学习, 协同过滤, 图卷积网络, 向量嵌入

Abstract: Collaborative filtering is a widely used recommendation algorithm whose core process is to learn the vector representation of users and goods.The Graph Convolution Network(GCN)-based collaborative filtering algorithm adds the correlation information of neighbor nodes in the vector embedding process, which further improves the algorithm's recommendation performance.However, the graph collaborative filtering algorithm tends to over smooth and only uses the adjacency matrix to expand the local structure.Furthermore, it does not mine the potential interaction patterns between nodes from the overall graph structure, making the interaction information source single.To overcome these limitations, a GCN-based Dual-Channel Collaborative Filtering Recommendation(DCCF) algorithm is proposed in this study.In the proposed algorithm, the vector embedding process is divided into local and global convolution channels to obtain different types of connection information.In the local convolution channel, the neighborhood nodes are directly located and a single-layer network structure is used to complete the calculation.Additionally, information aggregation is optimized to address the over smoothing problem.In the global convolution channel, the global interaction graph is constructed via clustering and participates in the information aggregation process, so as to mine the potential connections between nodes.Local and global information are combined to obtain a node vector representation containing different types of high-order relationships.Comparative experiments are conducted on three public datasets and the results show that the DCCF algorithm has a maximum increase of 2.8% and 5.0% in two indicators, namely, the normalized impairment cumulative gain and recall, respectively, compared with the best-performing model in the benchmark algorithm.

Key words: recommendation algorithm, deep learning, collaborative filtering, Graph Convolutional Network(GCN), vector embedding

中图分类号:

TP183

苗雨欣, 宋春花, 牛保宁, 康瑞雪. 双通道图协同过滤推荐算法[J]. 计算机工程, 2022, 48(8): 121-128.

MIAO Yuxin, SONG Chunhua, NIU Baoning, KANG Ruixue. Dual-Channel Graph Collaborative Filtering Recommendation Algorithm[J]. Computer Engineering, 2022, 48(8): 121-128.

https://www.ecice06.com/CN/Y2022/V48/I8/121

图/表 12

20220825091733

20220825091737

20220825091740

20220825091744

20220825091748

20220825091753

20220825091757

20220825091802

20220825091806

20220825091811

20220825091814

20220825091818

参考文献

[1] 黄立威, 江碧涛, 吕守业, 等.基于深度学习的推荐系统研究综述[J].计算机学报, 2018, 41(7):1619-1647. HUANG L W, JIANG B T, LÜS Y, et al.Survey on deep learning based recommender systems[J].Chinese Journal of Computers, 2018, 41(7):1619-1647.(in Chinese)
[2] 刘军, 杨军, 宋姗姗.基于用户购买意愿力的协同过滤推荐算法[J].吉林大学学报(理学版), 2021, 59(6):1432-1438. LIU J, YANG J, SONG S S.Collaborative filtering recommendation algorithm based on purchasing intention of users[J].Journal of Jilin University (Science Edition), 2021, 59(6):1432-1438.(in Chinese)
[3] 徐冰冰, 岑科廷, 黄俊杰, 等.图卷积神经网络综述[J].计算机学报, 2020, 43(5):755-780. XU B B, CEN K T, HUANG J J, et al.A survey on graph convolutional neural network[J].Chinese Journal of Computers, 2020, 43(5):755-780.(in Chinese)
[4] ZHOU J, CUI G Q, HU S D, et al.Graph neural networks:a review of methods and applications[J].AI Open, 2020, 1:57-81.
[5] WU S W, ZHANG W T, SUN F, et al.Graph neural networks in recommender systems:a survey[EB/OL].[2021-07-01].https://arxiv.org/abs/2011.02260.
[6] WANG X, HE X N, WANG M, et al.Neural graph collaborative filtering[C]//Proceedings of the 42nd International ACM SIGIR Conference on Research and Development in Information Retrieval.New York, USA:ACM Press, 2019:165-174.
[7] 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, USA:ACM Press, 2020:639-648.
[8] CHEN L, WU L, HONG R C, et al.Revisiting graph based collaborative filtering:a linear residual graph convolutional network approach[J].Proceedings of the AAAI Conference on Artificial Intelligence, 2020, 34(1):27-34.
[9] KIPF T N, WELLING M.Semi-supervised classification with graph convolutional networks[EB/OL].[2021-07-01].https://arxiv.org/abs/1609.02907.
[10] LI Q M, HAN Z C, WU X M.Deeper insights into graph convolutional networks for semi-supervised learning[EB/OL].[2021-07-01].https://arxiv.org/abs/1801.07606.
[11] HUANG W B, RONG Y, XU T Y, et al.Tackling over-smoothing for general graph convolutional networks[EB/OL].[2021-07-01].https://arxiv.org/abs/2008.09864.
[12] HECHTLINGER Y, CHAKRAVARTI P, QIN J.A generalization of convolutional neural networks to graph-structured data[EB/OL].[2021-07-01].https://arxiv.org/abs/1704.08165.
[13] HERLOCKER J L, KONSTAN J A, BORCHERS A, et al.An algorithmic framework for performing collaborative filtering[J].ACM SIGIR Forum, 2017, 51(2):227-234.
[14] SARWAR B, KARYPIS G, KONSTAN J, et al.Item-based collaborative filtering recommendation algorithms[C]//Proceedings of the 10th International Conference on World Wide Web.New York, USA:ACM Press, 2001:285-295.
[15] KOREN Y, BELL R, VOLINSKY C.Matrix factorization techniques for recommender systems[J].Computer, 2009, 42(8):30-37.
[16] RENDLE S, FREUDENTHALER C, GANTNER Z, et al.BPR:Bayesian personalized ranking from implicit feedback[EB/OL].[2021-07-01].https://arxiv.org/abs/1205.2618.
[17] 贾俊杰, 刘鹏涛, 陈旺虎.融合社交信息的矩阵分解改进推荐算法[J].计算机工程, 2021, 47(9):97-105. JIA J J, LIU P T, CHEN W H.Improved matrix factorization algorithm using social information for recommendation[J].Computer Engineering, 2021, 47(9):97-105.(in Chinese)
[18] HE X N, LIAO L Z, ZHANG H W, et al.Neural collaborative filtering[C]//Proceedings of the 26th International Conference on World Wide Web.New York, USA:ACM Press, 2017:173-182.
[19] HE X N, HE Z K, SONG J K, et al.NAIS:neural attentive item similarity model for recommendation[J].IEEE Transactions on Knowledge and Data Engineering, 2018, 30(12):2354-2366.
[20] BERG R, KIPF T N, WELLING M.Graph convolutional matrix completion[C]//Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery &Data Mining.New York, USA:ACM Press, 2018:1-9.
[21] JI S Y, FENG Y F, JI R R, et al.Dual channel hypergraph collaborative filtering[C]//Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery &Data Mining.New York, USA:ACM Press, 2020:2020-2029.
[22] FENG Y F, YOU H X, ZHANG Z Z, et al.Hypergraph neural networks[J].Proceedings of the AAAI Conference on Artificial Intelligence, 2019, 33:3558-3565.
[23] JIANG J, WEI Y, FENG Y, et al.Dynamic hypergraph neural networks[EB/OL].[2021-07-01].https://www.ijcai.org/Proceedings/2019/0366.pdf.
[24] XUE H J, DAI X, ZHANG J, et al.Deep matrix factorization models for recommender systems[EB/OL].[2021-07-01].https://www.ijcai.org/Proceedings/2017/0447.pdf.
[25] WANG X, WANG R J, SHI C, et al.Multi-component graph convolutional collaborative filtering[J].Proceedings of the AAAI Conference on Artificial Intelligence, 2020, 34(4):6267-6274.
[26] 康雁, 李涛, 李浩, 等.融合知识图谱与协同过滤的推荐模型[J].计算机工程, 2020, 46(12):73-79, 87. KANG Y, LI T, LI H, et al.Recommendation model fusing with knowledge graph and collaborative filtering[J].Computer Engineering, 2020, 46(12):73-79, 87.(in Chinese)

选择文件类型/文献管理软件名称

选择包含的内容