一种基于图神经网络的社会化推荐算法

doi:10.12068/j.issn.1005-3026.2024.01.002

摘要/Abstract

摘要：

现有的社会化推荐算法大多着眼于用户购买或点击等单一的交互行为，并未同时考虑收藏、浏览等多种不同的交互行为.而且当前的社会化推荐算法重点只关注推荐的准确性，忽略了推荐结果的可解释性.针对以上问题，提出了一种基于图神经网络的社会化推荐算法SRGN，将用户的社交关系和物品间客观存在的语义联系以特定的方式注入到算法架构中，并且利用消息传递的方式实现交互的多行为联合编码，从而提升推荐的准确性.此外，设计了可解释模块为推荐结果提供推荐的理由.在两个真实数据集上与其他8种算法进行对比实验，结果表明提出的算法在推荐性能和用户友好性上具有明显的优势.

Abstract:

Most existing social recommendation algorithms focus on the user’s single interaction such as purchase or click， but do not consider different interactions such as collection and browsing simultaneously. Moreover， current social recommendation algorithms only focus on the accuracy of recommendation， ignoring the interpretability of recommendation results. To solve the above problems， a social recommendation algorithm SRGN is proposed based on graph neural network， which injects the social relationships of users and the objectively existing semantic connections between items into the algorithm architecture in a specific way， and jointly encodes the interactive multi-behavior through message transmission， so as to improve the accuracy of recommendation. In addition， an explainable module is designed to provide reasons for the recommendation results. Compared with other eight algorithms on two real datasets， the results show that the proposed algorithm has obvious advantages in recommendation performance and user friendliness.

Key words: recommendation system, social recommendation, graph neural network, explainable recommendation, personalized recommendation

中图分类号:

TP 311

吕艳霞, 郝帅, 乔广通, 邢烨. 一种基于图神经网络的社会化推荐算法[J]. 东北大学学报（自然科学版）, 2024, 45(1): 10-17.

Yan-xia LYU, Shuai HAO, Guang-tong QIAO, Ye XING. A Social Recommendation Algorithm Based on Graph Neural Network[J]. Journal of Northeastern University(Natural Science), 2024, 45(1): 10-17.

图/表 7

参考文献 19

1	Liu C H， Wang X， Lu T，et al.Discrete social recommendation［J］.Proceedings of the AAAI Conference on Artificial Intelligence，2019，33（1）：208-215.
2	Lyu Y X， Zheng Y， Wei F N，et al.AICF：attention‐based item collaborative filtering［J］.Advanced Engineering Informatics，2020，44：101090.
3	丁来旭，刘洪娟.复杂网络上基于多维特征表示学习的推荐算法［J］.东北大学学报（自然科学版），2022，43（3）：359-367.
	Ding Lai‐xu， Liu Hong‐juan.Recommendation algorithm based on multi‐dimensional feature representation learning in complex networks［J］.Journal of Northeastern University （Natural Science），2022，43（3）：359-367.
4	Guo G B， Zhang J， Yorke‐Smith N.TrustSVD：collaborative filtering with both the explicit and implicit influence of user trust and of item ratings［J］.Proceedings of the Twenty‐Ninth AAAI Conference on Artificial Intelligence，2015，29（1）：123-129.
5	Wang X， He X N， Cao Y X，et al.Kgat：knowledge graph attention network for recommendation［C］//Proceedings of the KDD Conference on Machine Learning.Anchorage：KDD，2019：950–958.
6	吴博，梁循，张树森，等.图神经网络前沿进展与应用［J］.计算机学报，2022，45（1）：35-68.
	Wu Bo， Liang Xun， Zhang Shu‐sen，et al.Advances and applications in graph neural network［J］.Chinese Journal of Computers，2022，45（1）：35-68.
7	Chen C， Zhang M， Liu Y Q，et al.Social attentional memory network：modeling aspect‐and friend‐level differences in recommendation［C］//Proceedings of the Twelfth ACM International Conference on Web Search and Data Mining.Melbourne，2019：177-185.
8	Chen C， Zhang M， Wang C Y，et al.An efficient adaptive transfer neural network for social‐aware recommendation［C］//Proceedings of the 42nd International ACM SIGIR Conference on Research and Development in Information Retrieval.Paris，2019：225-234.
9	Wu Q T， Zhang H R， Gao X F，et al.Dual graph attention networks for deep latent representation of multifaceted social effects in recommender systems［C］//The World Wide Web Conference.San Francisco，2019：2091-2102.
10	Fan W Q， Ma Y， Li Q，et al.Graph neural networks for social recommendation［C］//The World Wide Web Conference.San Francisco，2019：417-426.
11	Ying R， He R N， Chen K F，et al.Graph convolutional neural networks for web‐scale recommender systems［C］//Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining.London，2018：974-983.
12	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.Paris，2019：165-174.
13	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.NewYork，2020：639-648.
14	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.
15	Peake G， Wang J.Explanation mining：post hoc interpretability of latent factor models for recommendation systems［C］//Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining.London，2018：2060-2069.
16	朱海萍，赵成成，刘启东，等.基于互惠性约束的可解释就业推荐方法［J］.计算机研究与发展，2021，58（12）：2660-2672.
	Zhu Hai‐ping， Zhao Cheng‐cheng， Liu Qi‐dong，et al.Reciprocal‐constrained interpretable job recommendation［J］.Journal of Computer Research and Development，2021，58（12）：2660-2672.
17	Yang B， Lei Y， Liu J M，et al.Social collaborative filtering by trust［J］.IEEE Transactions on Pattern Analysis and Machine Intelligence，2017，39（8）：1633-1647.
18	Huang C， Xu H C， Xu Y，et al.Knowledge‐aware coupled graph neural network for social recommendation［J］.Proceedings of the AAAI Conference on Artificial Intelligence，2021，35（5）：4115-4122.
19	Song W P， Xiao Z P， Wang Y F，et al.Session‐based social recommendation via dynamic graph attention networks［C］//Proceedings of the Twelfth ACM International Conference on Web Search and Data Mining.Melbourne，2019：555-563.

社交关系	Epinions	Yelp
用户数量	18 081	43 043
物品数量	251 722	66 576
用户-物品交互数量	715 821	283 512
社会关系数量	590 641	549 451
物品关系数量	6 096 106	1 847 060

社交关系	Epinions	Yelp
用户数量	18 081	43 043
物品数量	251 722	66 576
用户-物品交互数量	715 821	283 512
社会关系数量	590 641	549 451
物品关系数量	6 096 106	1 847 060

算法	Top-K@5		Top-K@10		Top-K@15
算法	HR	NDCG	HR	NDCG	HR	NDCG
TrustMF	0.631 3	0.462 5	0.758 5	0.512 8	0.863 1	0.533 4
SAMN	0.599 7	0.436 7	0.751 8	0.486 5	0.827 3	0.507
DGRec	0.611 7	0.444 8	0.766 5	0.495 5	0.839 8	0.514 7
EATNN	0.626 0	0.455 5	0.771 7	0.506 9	0.841 4	0.525 4
KGAT	0.639 7	0.467 5	0.772 2	0.511 4	0.854 2	0.532 9
GraphRec	0.623 4	0.450 6	0.760 8	0.494 4	0.834 2	0.513 6
DANSER	0.630 6	0.462 7	0.774 3	0.508 5	0.835 8	0.524 7
KCGN	0.659 5	0.487 6	0.802 6	0.530 8	0.868 1	0.542 6
SRGN	0.681 8	0.510 4	0.821 3	0.555 7	0.882 8	0.569 8

算法	Top-K@5		Top-K@10		Top-K@15
算法	HR	NDCG	HR	NDCG	HR	NDCG
TrustMF	0.631 3	0.462 5	0.758 5	0.512 8	0.863 1	0.533 4
SAMN	0.599 7	0.436 7	0.751 8	0.486 5	0.827 3	0.507
DGRec	0.611 7	0.444 8	0.766 5	0.495 5	0.839 8	0.514 7
EATNN	0.626 0	0.455 5	0.771 7	0.506 9	0.841 4	0.525 4
KGAT	0.639 7	0.467 5	0.772 2	0.511 4	0.854 2	0.532 9
GraphRec	0.623 4	0.450 6	0.760 8	0.494 4	0.834 2	0.513 6
DANSER	0.630 6	0.462 7	0.774 3	0.508 5	0.835 8	0.524 7
KCGN	0.659 5	0.487 6	0.802 6	0.530 8	0.868 1	0.542 6
SRGN	0.681 8	0.510 4	0.821 3	0.555 7	0.882 8	0.569 8

[1]	刘海博，冯时，于戈. 一种基于旁信息增强的协同过滤自动编码器模型[J]. 东北大学学报:自然科学版, 2019, 40(11): 1527-1532.
[2]	赵晓煜;黄小原;曹忠鹏;. 基于顾客交易数据的协同过滤推荐方法[J]. 东北大学学报(自然科学版), 2009, 30(12): 1792-1795.
[3]	鲍玉斌;王大玲;于戈. 关联规则和聚类分析在个性化推荐中的应用[J]. 东北大学学报(自然科学版), 2003, 24(12): 1149-1152.