基于双图混合随机游走的社会化推荐模型

doi:10.12263/DZXB.20210504

电子学报

• •

基于双图混合随机游走的社会化推荐模型

曹阳¹^,², 高旻¹^,²(), 余俊良³, 范琪琳¹^,², 荣文戈⁴, 文俊浩¹^,²

^1.信息物理社会可信服务计算教育部重点实验室（重庆大学），重庆 400044
^2.重庆大学大数据与软件学院，重庆 400044
^3.昆士兰大学信息技术与电气工程学院，昆士兰州 4072 澳大利亚
^4.北京航空航天大学计算机学院，北京 10019

收稿日期:2021-04-20 修回日期:2021-06-28 出版日期:2022-07-18
- 通讯作者:
- 高旻
- 作者简介:
- 曹阳男，1996年生，现为重庆大学大数据与软件学院在读硕士研究生.主要研究方向为机器学习、推荐系统等.E-mail: cqucaoy.cqu.edu.cn
  高旻（通讯作者）女，1980年生，工学博士，重庆大学大数据与软件学院副教授、硕士生导师.主要研究包括推荐系统、异常检测、社会媒体挖掘. E-mail:gaomin@cqu.deu.cn
  余俊良男，1990年生，在重庆大学大数据与软件学院获得学士学位和硕士学位.现为澳大利亚昆士兰大学信息技术与电气工程学院在读博士研究生.主要研究方向为推荐系统和异常检测等.E-mail: jl.yu@uq.edu.au
  范琪琳女，1989年生，工学博士，重庆大学大数据与软件学院讲师、硕士生导师.2011年在四川大学软件工程学院获得学士学位，2017年在清华大学计算机科学与技术学院获得博士学位.研究兴趣包括网络优化、移动边缘计算与缓存、网络虚拟化和机器学习.E-mail: fanqilin@cqu.edu.cn
  荣文戈男，1975年生，工学博士，北京航空航天大学计算机学院教授、博士生导师.主要研究方向为机器学习、自然语言处理、数据挖掘和信息系统等.E-mail: w.rong@buaa.edu.cn
  文俊浩男，1969年生，工学博士，重庆大学软件学院教授.主要研究方向为计算智能与推荐系统.E-mail: jhwen@cqu.deu.cn
- 基金资助:
- 国家自然科学基金 (62176028); 重庆市自然科学基金面上项目 (cstc2020jcyj-msxmX0690); 中央高校基本科研业务费项目 (2020CDJ-LHZZ-039); 重庆市留学人员创业创新支持计划 (cx2020097); 重庆市技术创新与应用示范专项产业类重点研发项目 (cstc2019jscx-mbdxX0008); 重庆市技术创新与应用发展专项 (cstc2019jscx-zdztzxX0031)

Bi-Graph Mix-random Walk Based Social Recommendation Model

CAO Yang¹^,², GAO Min¹^,²(), YU Jun-liang³, FAN Qi-lin¹^,², RONG Wen-ge⁴, WEN Jun-hao¹^,²

^1.Key Laboratory of Dependable Service Computing in Cyber Physical Society （Chongqing University），Ministry of Education，Chongqing 400044，China
^2.School of Big Data & Software Engineering，Chongqing University，Chongqing 400044，China
^3.School of Information Technology and Electrical Engineering，University of Queensland，QLD 4072，AUS
^4.School of Computer Science and Engineering，Beihang University，Beijing 100191，China

Received:2021-04-20 Revised:2021-06-28 Online:2022-07-18
- Corresponding author:
- GAO Min

摘要/Abstract

摘要：

近年来，可以有效缓解数据稀疏和冷启动问题的社会化推荐受到了研究者和业界的关注.社会化推荐利用显式或隐式社交关系作为辅助信息，提升了推荐性能.然而，目前的社会化推荐模型通常采用普通图描述社交关系.普通图中的边常描述为成对节点的关系，这种方法适合描述显式关系，但难以描述复杂的隐式关系，如购买过同一商品的多个用户之间的集合关系，因此难以学习到准确的节点表示，影响推荐的性能.针对此问题，本论文结合超图和普通图，提出基于双图混合随机游走的推荐模型（BG-Rec）.构建超图描述复杂的隐式关系，同时用普通图描述显式的社交关系，并在两种图上定义混合随机游走策略，生成结合隐式关系和显式关系的游走节点序列，学习更准确的节点嵌入表示.根据用户评分的高低，构建了正反馈超图和负反馈超图，考虑更细粒度的朋友关系，以识别可靠的朋友.融合可靠朋友的偏好和后验概率最大化优化物品个性化排序.三个公开数据集的大量实验表明了BG-Rec在推荐性能上的优越性，冷启动和消融实验表明了其在缓解冷启动问题的有效性和超图建模的合理性.

关键词: 推荐系统, 社交关系, 随机游走, 超图, 个性化排序

Abstract:

In recent years, social recommendation approaches have attracted attention because they can effectively improve the recommendation quality when user-item interaction data is sparse. Explicit and implicit social relations, as auxiliary information, are used to improve the recommendation quality. However, social relations are represented by simple graphs in existing models. The nature of edges connecting pair-wise nodes in simple graphs makes it suitable for describing explicit relations. Still, it is incapable of modeling complex implicit relations, such as the collective relation between multiple users who have purchased the same product. Therefore, it isn't easy to learn the node representation accurately, only based on simple graphs, which even affects the recommender's performance. In this paper, we propose a recommendation model (BG-Rec) based on a Bi-graph hybrid random walk to overcome this problem, which combines hypergraph and graph. We construct a hypergraph and a simple graph to depict complex implicit relations and explicit social relations separately. Next, the mixed random walk strategy (MixRandom) is used to generate node sequences that combine implicit and explicit relations. Furthermore, node sequences are used for learning more accurate representations of nodes. Then, positive feedback hypergraph and negative feedback hypergraph are constructed based on user ratings, so that more fine-grained friend relations can be considered to identify reliable friends. Finally, the personalized ranking of items is optimized by considering the preferences of reliable friends and the maximization of the posterior probability. Experiments on three public datasets show the superiority of BG-Rec in recommendation performance. The cold-start study and ablation study validates the effectiveness of alleviating the cold-start problem and rationality of hypergraph modeling.

Key words: recommendation systems, social relations, random walk, hypergraph, personalized ranking

中图分类号:

TP311

曹阳, 高旻, 余俊良, 范琪琳, 荣文戈, 文俊浩. 基于双图混合随机游走的社会化推荐模型[J]. 电子学报, DOI: 10.12263/DZXB.20210504.

CAO Yang, GAO Min, YU Jun-liang, FAN Qi-lin, RONG Wen-ge, WEN Jun-hao. Bi-Graph Mix-random Walk Based Social Recommendation Model[J]. Acta Electronica Sinica, DOI: 10.12263/DZXB.20210504.

图/表 13

参考文献 31

1	YU X, JIANG F, DU J, et al. A cross-domain collaborative filtering algorithm with expanding user and item features via the latent factor space of auxiliary domains [J]. Pattern Recognition, 2019, 94(1): 96-109.
2	QI L, WANG X, XU X, et al. Privacy-aware cross-platform service recommendation based on enhanced locality-sensitive hashing [J]. IEEE Transactions on Network Science and Engineering, 2021, 8(2): 1145-1153.
3	李琳,唐守廉.基于多层注意力表示的音乐推荐模型 [J].电子学报, 2020, 48(9): 1672-1679.
	LI L, TANG S L. Hierarchical attention representation model for music recommendation[J]. Acta Electronica Sinica, 2020, 48(9): 1672-1679. (In Chinese)
4	郭景峰,朱晓松,李爽.直播电视推荐系统的评分预测算法研究 [J]. 电子学报, 2020, 48(9): 1735-1740.
	GUO J F, ZHU X S, LI S. Research on rating prediction algorithm of live TV recommender systems[J]. Acta Electronica Sinica, 2020, 48(9): 1735-1740. (in Chinese)
5	CIALDINI R B, GOLDSTEIN N J. Social influence: Compliance and conformity [J]. Annu Rev Psychol, 2004, 55(1): 591-621.
6	SINHA R R, SWEARINGEN K. Comparing recommendations made by online systems and friends[C]//Proceedings of the Second DELOS Network of Excellence Workshop on Personalisation and Recommender Systems in Digital Libraries. Sophia Antipolis: ERCIM, 2001: 1-6.
7	DOU K, GUO B, KUANG L. A privacy-preserving multimedia recommendation in the context of social network based on weighted noise injection [J]. Multimedia Tools and Applications, 2019, 78(19): 26907-26926.
8	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 Conference on Information and Knowledge Management. New York: ACM, 2014: 261-270.
9	MA H, KING I, LYU M R. Learning to recommend with social trust ensemble[C]//Proceedings of the 32nd International ACM SIGIR Conference on Research and Development in Information Retrieval. New York: ACM, 2009: 203-210.
10	KANG M, BI Y, WU Z, et al. A heterogeneous conversational recommender system for financial products[C]//Proceedings of the 28th ACM International Conference on Information and Knowledge Management. New York: ACM, 2019: 26-30.
11	ZHAO H, ZHOU Y, SONG Y, et al. Motif enhanced recommendation over heterogeneous information network[C]//Proceedings of the 28th ACM International Conference on Information and Knowledge Management. New York: ACM, 2019: 2189-2192.
12	YU J, GAO M, LI J, et al. Adaptive implicit friends identification over heterogeneous network for social recommendation[C]//Proceedings of the 27th ACM International Conference on Information and Knowledge Management. New York: ACM, 2018: 357-366.
13	YU J, YIN H, LI J, et al. Self-supervised multi-channel hypergraph convolutional network for social recommendation[C]//The Web Conference 2021. New York: ACM, 2021: 413-424.
14	JI S, FENG Y, JI R, et al. Dual channel hypergraph collaborative filtering[C]//Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. New York: ACM, 2020: 2020-2029.
15	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. New York: ACM, 2008: 931-940.
16	YANG B, LEI Y, LIU J, et al. Social collaborative filtering by trust [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016, 39(8): 1633-1647.
17	CHEN J, FENG Y, ESTER M, et al. Modeling users' exposure with social knowledge influence and consumption influence for recommendation[C]//Proceedings of the 27th ACM International Conference on Information and Knowledge Management. New York: ACM, 2018: 953-962.
18	RENDLE S, FREUDENTHALER C, GANTNER Z, et al. BPR: Bayesian personalized ranking from implicit feedback[C]//In Proceedings of the Twenty-fifth Conference on Uncertainty in Artificial Intelligence. Arlington: AUAI Press, 2009: 452-461
19	TANG J, HU X, LIU H. Social recommendation: a review [J]. Social Network Analysis and Mining, 2013, 3(4): 1113-1133.
20	WANG X, HOI S C H, ESTER M, et al. Learning personalized preference of strong and weak ties for social recommendation[C]//Proceedings of the 26th International Conference on World Wide Web. Geneva: IW 3C2, 2017: 1601-1610.
21	GOYAL P, FERRARA E. Graph embedding techniques, applications, and performance: A survey [J]. Knowledge-Based Systems, 2018, 151(1): 78-94.
22	FAN W, MA Y, LI Q, et al. Graph neural networks for social recommendation[C]//Proceedings of the Web Conference 2019. New York: ACM, 2019: 417-426.
23	WU L, SUN P, FU Y, 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: ACM, 2019: 235-244.
24	YU J, GAO M, YIN H, et al. Generating reliable friends via adversarial training to improve social recommendation[C]//2019 IEEE International Conference on Data Mining. New York: IEEE, 2019: 768-777.
25	YU J, YIN H, LI J, et al. Enhance social recommendation with adversarial graph convolutional networks [EB/OL]. (2020-10-23)[2021-06-28]. .
26	CHITRA U, RAPHAEL B J. Random walks on hypergraphs with edge-dependent vertex weights[C]//Proceedings of the 36th International Conference on Machine Learning. Long Beach: IMLS, 2019: 2002-2011.
27	FROLOV E, OSELEDETS I. Fifty shades of ratings: How to benefit from a negative feedback in top-N recommendations tasks[C]//Proceedings of the 10th ACM Conference on Recommender Systems. New York: ACM, 2016: 91-98.
28	DONG Y, CHAWLA N V, SWAMI A. Metapath2vec: Scalable representation learning for heterogeneous networks[C]//Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York: ACM, 2017: 135-144.
29	MIKOLOV T, SUTSKEVER I, CHEN K, et al. Distributed representations of words and phrases and their compositionality [J]. Advances in Neural Information Processing Systems, 2013, 26(1): 3111-3119.
30	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. New York: ACM, 2020: 639-648.
31	WU J, WANG X, FENG F, et al. Self-supervised graph learning for recommendation[C]//Proceedings of the 44rd International ACM SIGIR Conference on Research and Development in Information Retrieval. New York: ACM, 2021: 639-648.

Dataset	Users	Items	Feedbacks	Density	Relations
LastFM	1 892	17 632	92 834	0.28%	25 434
Douban	2 831	22 129	606 467	0.97%	35 624
Epinions	15 064	23 575	356 959	0.10%	285 659

Dataset	Users	Items	Feedbacks	Density	Relations
LastFM	1 892	17 632	92 834	0.28%	25 434
Douban	2 831	22 129	606 467	0.97%	35 624
Epinions	15 064	23 575	356 959	0.10%	285 659

Dataset	Metric	BPR	SBPR	RSGAN	IF-BPR	LightGCN	SGL	BG-Rec⁺	Improve
LastFM	Prec@10	0.042 6	0.087	0.081 7	0.102 1	0.096 7	0.077	0.105 0	2.84%
	Rec@10	0.064 8	0.130 9	0.122 4	0.152 3	0.145	0.115 7	0.157 7	3.55%
	F1@10	0.051 4	0.104 5	0.097 9	0.122 2	0.116	0.092 5	0.126 1	3.19%
	MAP@10	0.025 8	0.058 7	0.054 5	0.070 1	0.064 5	0.052 6	0.072 6	3.57%
	NDCG@10	0.060 2	0.125 7	0.118 6	0.148 2	0.138 3	0.113 4	0.152 5	2.90%
Douban	Prec@10	0.123 6	0.158 8	0.173	0.175 7	0.125 7	0.146 2	0.182 0	3.59%
	Rec@10	0.041 3	0.055 4	0.057 5	0.058 3	0.047 6	0.048 4	0.061 5	5.49%
	F1@10	0.061 9	0.082 2	0.086 3	0.087 5	0.069	0.072 7	0.092 0	5.14%
	MAP@10	0.071 1	0.097 2	0.108 3	0.109 7	0.067 2	0.084	0.115 1	4.92%
	NDCG@10	0.140 8	0.181 2	0.195 5	0.197 4	0.137 8	0.162 1	0.205 4	4.05%
	NDCG@10	0.140 8	0.181 2	0.195 5	0.197 4	0.137 8	0.162 1	0.205 4	4.05%
Epinions	Prec@10	0.011	0.014 5	0.018 6	0.015 7	0.016 7	0.019 2	0.020 0	4.17%
	Rec@10	0.025 1	0.030 9	0.041 5	0.034 2	0.036 9	0.042 4	0.044 0	3.77%
	F1@10	0.015 3	0.019 7	0.025 7	0.021 5	0.023	0.026 4	0.027 5	4.17%
	MAP@10	0.009	0.012 2	0.016 6	0.015 1	0.014 8	0.018 6	0.019 1	2.69%
	NDCG@10	0.019 4	0.025 3	0.033 5	0.029 2	0.030 1	0.035 9	0.037 3	3.90%
	NDCG@10	0.019 4	0.025 3	0.033 5	0.029 2	0.030 1	0.035 9	0.037 3	3.90%

Dataset	Metric	BPR	SBPR	RSGAN	IF-BPR	LightGCN	SGL	BG-Rec⁺	Improve
LastFM	Prec@10	0.042 6	0.087	0.081 7	0.102 1	0.096 7	0.077	0.105 0	2.84%
	Rec@10	0.064 8	0.130 9	0.122 4	0.152 3	0.145	0.115 7	0.157 7	3.55%
	F1@10	0.051 4	0.104 5	0.097 9	0.122 2	0.116	0.092 5	0.126 1	3.19%
	MAP@10	0.025 8	0.058 7	0.054 5	0.070 1	0.064 5	0.052 6	0.072 6	3.57%
	NDCG@10	0.060 2	0.125 7	0.118 6	0.148 2	0.138 3	0.113 4	0.152 5	2.90%
Douban	Prec@10	0.123 6	0.158 8	0.173	0.175 7	0.125 7	0.146 2	0.182 0	3.59%
	Rec@10	0.041 3	0.055 4	0.057 5	0.058 3	0.047 6	0.048 4	0.061 5	5.49%
	F1@10	0.061 9	0.082 2	0.086 3	0.087 5	0.069	0.072 7	0.092 0	5.14%
	MAP@10	0.071 1	0.097 2	0.108 3	0.109 7	0.067 2	0.084	0.115 1	4.92%
	NDCG@10	0.140 8	0.181 2	0.195 5	0.197 4	0.137 8	0.162 1	0.205 4	4.05%
	NDCG@10	0.140 8	0.181 2	0.195 5	0.197 4	0.137 8	0.162 1	0.205 4	4.05%
Epinions	Prec@10	0.011	0.014 5	0.018 6	0.015 7	0.016 7	0.019 2	0.020 0	4.17%
	Rec@10	0.025 1	0.030 9	0.041 5	0.034 2	0.036 9	0.042 4	0.044 0	3.77%
	F1@10	0.015 3	0.019 7	0.025 7	0.021 5	0.023	0.026 4	0.027 5	4.17%
	MAP@10	0.009	0.012 2	0.016 6	0.015 1	0.014 8	0.018 6	0.019 1	2.69%
	NDCG@10	0.019 4	0.025 3	0.033 5	0.029 2	0.030 1	0.035 9	0.037 3	3.90%
	NDCG@10	0.019 4	0.025 3	0.033 5	0.029 2	0.030 1	0.035 9	0.037 3	3.90%

Model	Component	Dataset	Prec@10	Rec@10	F1@10	MAP@10	NDCG@10
BG-Rec^w/oH	Graph√	LastFM	0.102 1	0.152 3	0.122 2	0.070 1	0.148 2
		Douban	0.175 7	0.058 3	0.087 5	0.109 7	0.197 4
		Epinions	0.015 7	0.034 2	0.021 5	0.015 1	0.029 2
BG-Rec^w/oG	Hypergraph√	LastFM	0.100 0	0.150 9	0.120 3	0.068 5	0.144 5
		Douban	0.152 7	0.055 9	0.081 9	0.092 7	0.173 4
		Epinions	0.010 2	0.023 1	0.014 2	0.009 2	0.018 6
BG-Rec+	Graph√	LastFM	0.105 0	0.157 7	0.126 1	0.072 6	0.152 5
	Graph√	Douban	0.182 0	0.061 5	0.092 0	0.115 1	0.205 4
	Hypergraph√	Epinions	0.020 0	0.044 0	0.027 5	0.019 1	0.037 3

基于双图混合随机游走的社会化推荐模型

Bi-Graph Mix-random Walk Based Social Recommendation Model

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 13

参考文献 31

相关文章 15

编辑推荐

Metrics

[1]	王雪松, 张翰林, 程玉虎. 基于自编码器和超图的半监督宽度学习系统[J]. 电子学报, 2022, 50(3): 533-539.
[2]	张帅, 高旻, 文俊浩, 熊庆宇, 唐旭. 基于自监督学习的去流行度偏差推荐方法[J]. 电子学报, 2022, 50(10): 2361-2371.
[3]	李青青, 马慧芳, 李举, 李志欣. 属性网络中相似性增强随机游走的多社区搜索[J]. 电子学报, 2021, 49(11): 2096-2100.
[4]	郑孝遥, 罗永龙, 汪祥舜, 孙丽萍, 陈付龙, 胡桂银, 汪小寒. 基于位置服务的分布式差分隐私推荐方法研究[J]. 电子学报, 2021, 49(1): 99-110.
[5]	郭景峰, 朱晓松, 李爽. 直播电视推荐系统的评分预测算法研究[J]. 电子学报, 2020, 48(9): 1735-1740.
[6]	黄名选, 蒋曹清. 基于项权值排序挖掘的跨语言查询扩展[J]. 电子学报, 2020, 48(3): 568-576.
[7]	王雄, 董一鸿, 潘剑飞, 陈华辉, 钱江波. 基于编辑行为码的图概要技术[J]. 电子学报, 2020, 48(12): 2434-2443.
[8]	刘露, 胡封晔, 牛亮, 彭涛. 异质网络中基于节点影响力的相似度度量方法[J]. 电子学报, 2019, 47(9): 1929-1936.
[9]	李敏, 王晓桐, 罗慧敏, 孟祥茂, 王建新. 随机游走技术在网络生物学中的研究进展[J]. 电子学报, 2018, 46(8): 2035-2048.
[10]	肖云鹏, 孙华超, 戴天骥, 李茜, 李暾. 一种基于云模型的社交网络推荐系统评分预测方法[J]. 电子学报, 2018, 46(7): 1762-1767.
[11]	马慧芳, 刘芳, 夏琴, 郝占军. 基于加权超图随机游走的文献关键词提取算法[J]. 电子学报, 2018, 46(6): 1410-1414.
[12]	肖云鹏, 刘晏驰, 刘红, 刘媛妮. 一种基于动态角色标识和张量分解的推荐模型[J]. 电子学报, 2018, 46(3): 569-574.
[13]	袁红星, 安鹏, 吴少群, 郑悠. 具有容错能力的L₁最优化半自动2D转3D[J]. 电子学报, 2018, 46(2): 447-455.
[14]	黄名选, 蒋曹清. 基于完全加权正负关联模式挖掘的越-英跨语言查询译后扩展[J]. 电子学报, 2018, 46(12): 3029-3036.
[15]	俞东进, 陈聪, 吴建华, 陈耀旺. 基于隐式反馈数据的个性化游戏推荐[J]. 电子学报, 2018, 46(11): 2626-2632.