融合项目影响力的图神经网络会话推荐模型

doi:10.11772/j.issn.1001-9081.2022121812

《计算机应用》唯一官方网站 ›› 2023, Vol. 43 ›› Issue (12): 3689-3696.DOI: 10.11772/j.issn.1001-9081.2022121812

融合项目影响力的图神经网络会话推荐模型

孙轩宇¹, 史艳翠¹^,²()

^1.天津科技大学人工智能学院，天津 300457
^2.国家开放大学数字化学习技术集成与应用教育部工程研究中心，北京 100039

收稿日期:2022-12-07 修回日期:2023-03-05 接受日期:2023-03-07 发布日期:2023-12-11 出版日期:2023-12-10
通讯作者: 史艳翠
作者简介:孙轩宇（1998—），男，江苏南京人，硕士研究生，CCF会员，主要研究方向：深度学习、推荐系统；
基金资助:
天津市教委理工类基本科研业务费项目(2018KJ105);数字化学习技术集成与应用教育部工程研究中心创新基金资助项目(1221025)

Session-based recommendation model by graph neural network fused with item influence

Xuanyu SUN¹, Yancui SHI¹^,²()

^1.College of Artificial Intelligence，Tianjin University of Science & Technology，Tianjin 300457，China
^2.Engineering Research Center of Integration and Application of Digital Learning Technology，Ministry of Education，The Open University of China，Beijing 100039，China

Received:2022-12-07 Revised:2023-03-05 Accepted:2023-03-07 Online:2023-12-11 Published:2023-12-10
Contact: Yancui SHI
About author:SUN Xuanyu， born in 1998， M. S. candidate. His research interests include deep learning， recommender system.
Supported by:
Basic Research Business Fee Project for Science and Engineering of Tianjin Education Commission(2018KJ105);Innovation Fund of Engineering Research Center of Ministry of Education for Integration and Application of Digital Learning Technology(1221025)

摘要/Abstract

摘要：

针对现有的会话推荐模型难以显式地表示项目对推荐结果的影响的问题，提出一种融合项目影响力的图神经网络会话推荐模型（SR-II）。首先，提出一种新的边权重计算方法，将计算结果作为图结构中转移关系的影响力权重，并用图神经网络（GNN）的影响力图门控层提取该图的特征；其次，提出改进的捷径图连接有关联的项目，有效捕获远程依赖，丰富图结构所能表达的信息，并通过注意力机制的捷径图注意力层提取该图的特征；最后，通过结合上述两层，构建推荐模型。在Diginetica和Gowalla数据集上的实验结果中，SR-II的HR@20最高达到53.12%，MRR@20最高达到25.79%。在Diginetica数据集上，相较于同一表征空间下基于训练模型的会话推荐（CORE-trm），SR-II在HR@20上提升了1.10%，在MRR@20上提升了1.21%。在Gowalla数据集上，相较于基于会话的自注意网络推荐（SR-SAN），SR-II在HR@20上提升了1.73%；相较于基于无损边缘保留聚合和捷径图注意力的推荐（LESSR）模型，SR-II在MRR@20上提升了1.14%。实验结果表明SR-II的推荐效果优于对比模型，具有更高的推荐精度。

关键词: 会话推荐, 推荐系统, 图神经网络, 注意力机制, 会话图

Abstract:

Aiming at the problem that it is difficult for the existing session-based recommendation models to explicitly express the influence of items on the recommendation results， a Session-based Recommendation model by graph neural network fused with Item Influence （SR-II） was proposed. Firstly， a new edge weight calculation method was proposed to construct a graph structure， in which the calculated result was used as the influence weight of the transition relationship in the graph， and the features of the graph were extracted through the influence graph gated layer by using Graph Neural Network （GNN）. Then， an improved shortcut graph was proposed to connect related items， effectively capture long-range dependencies， and enrich the information expressed by the graph structure； and the features of the graph were extracted through the shortcut graph attention layer by using the attention mechanism. Finally， a recommendation model was constructed by combining the above two layers. In the experimental results on Diginetica and Gowalla datasets， the highest HR@20 of SR-II is reaching 53.12%， and the highest MRR@20 of SR-II is reaching 25.79%. On Diginetica dataset， compared with CORE-trm （simple and effective session-based recommendation within COnsistent REpresentation space-transformer）， SR-II has the HR@20 improved by 1.10% ，and the MRR@20 improved by 1.21%； On Gowalla dataset， compared with SR-SAN（Session-based Recommendation with Self-Attention Networks）， SR-II has the HR@20 improved by 1.73%.Compared with the recommendation model called LESSR （Lossless Edge-order preserving aggregation and Shortcut graph attention for Session-based Recommendation）， SR-II has the MRR@20 improved by 1.14%. The experimental results show that the performance of SR-II is better than that of the comparison models， and SR-II has a higher recommendation accuracy.

Key words: session-based recommendation, recommender system, Graph Neural Network (GNN), attention mechanism, session graph

中图分类号:

TP391.4

孙轩宇, 史艳翠. 融合项目影响力的图神经网络会话推荐模型[J]. 计算机应用, 2023, 43(12): 3689-3696.

Xuanyu SUN, Yancui SHI. Session-based recommendation model by graph neural network fused with item influence[J]. Journal of Computer Applications, 2023, 43(12): 3689-3696.

图/表 10

图1 本文模型的结构

Fig.1 Structure of proposed model

图2 会话图

Fig.2 Session graph

图3 构建影响力图

Fig. 3 Constructing influence graph

图4 构建捷径图

Fig. 4 Constructing shortcut graph

表1 实验数据集的统计信息

Tab. 1 Statistical information of experiment datasets

数据集	点击数	会话数	项目数	平均长度
Diginetica	981 620	777 029	42 596	4.80
Gowalla	1 122 788	830 893	29 510	3.85

表2 不同模型在两个数据集上的HR@20和MRR@20比较 ( %)

Tab. 2 Comparison of HR@20 and MRR@20 among different models on two datasets

模型	Diginetica		Gowalla
模型	HR@20	MRR@20	HR@20	MRR@20
Item-KNN	39.51	11.32	38.60	16.66
FPMC	28.50	7.67	29.91	11.45
NARM	49.80	16.54	50.07	23.93
SR-GNN	50.81	17.31	50.32	24.25
GC-SAN	50.90	17.59	50.68	24.67
LESSR	51.75	18.17	51.37	25.50
SR-SAN	51.94	17.55	51.52	24.83
CORE-trm	52.54	18.20	51.19	24.66
SR-II-base	52.71	18.23	52.00	25.35
SR-II	53.12	18.42	52.41	25.79

图5 3个模型的表现

Fig.5 Performance of three models

图6 不同θ下SR-II的表现

Fig. 6 Performance of SR-II with different θ

图7 SR-II的表现

Fig. 7 Performance of SR-II

表3 不同模型的时间复杂度和平均训练时间对比

Tab. 3 Comparisons of time complexity and average training time among different models

模型	时间复杂度	平均训练时间/s
SR-GNN	$O (V 2 d)$	437.69
GC-SAN	$O (V 2 d + V d 2)$	627.10
LESSR	$O (V 2 d + ɛ d)$	412.24
SR-SAN	$O (V 2 d + V d 2)$	441.26
CORE-trm	$O (V 2 d)$	201.35
SR-II-base	$O (V 2 d + ɛ d)$	527.14
SR-II	$O (V 2 d + ɛ d + ɛ 2)$	3 057.72

表3 不同模型的时间复杂度和平均训练时间对比

Tab. 3 Comparisons of time complexity and average training time among different models

模型	时间复杂度	平均训练时间/s
SR-GNN	$O (V 2 d)$	437.69
GC-SAN	$O (V 2 d + V d 2)$	627.10
LESSR	$O (V 2 d + ɛ d)$	412.24
SR-SAN	$O (V 2 d + V d 2)$	441.26
CORE-trm	$O (V 2 d)$	201.35
SR-II-base	$O (V 2 d + ɛ d)$	527.14
SR-II	$O (V 2 d + ɛ d + ɛ 2)$	3 057.72

参考文献 31

1	SHI L， ZHANG Y， CHENG J， et al. Skeleton-based action recognition with directed graph neural networks ［C］// Proceedings of the 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2019： 7912-7921. 10.1109/cvpr.2019.00810
2	GOMES J， RAMSUNDAR B， FEINBERG E N， et al. Atomic convolutional networks for predicting protein-ligand binding affinity［EB/OL］. ［2022-05-10］. .
3	MA T， XIAO C， ZHOU J， et al. Drug similarity integration through attentive multi-view graph auto-encoders［C］// Proceedings of the 27th International Joint Conference on Artificial Intelligence. California： IJCAI， 2018： 3477-3483. 10.24963/ijcai.2018/483
4	YU B， YIN H， ZHU Z. Spatio-temporal graph convolutional networks： a deep learning framework for traffic forecasting［C］// Proceedings of the 27th International Joint Conference on Artificial Intelligence. California： IJCAI， 2018： 3634-3640. 10.24963/ijcai.2018/505
5	KOSARAJU V， SADEGHIAN A， MARTÍN-MARTÍN R， et al. Social-bigat： multimodal trajectory forecasting using bicycle-gan and graph attention networks［C］// Proceedings of the 33rd International Conference on Neural Information Processing Systems. ‎［S.l.］： NIPS， 2019： 137-146.
6	BRUNA J， ZAREMBA W， SZLAM A， et al. Spectral networks and locally connected networks on graphs［EB/OL］. （2014-04-21）［2022-07-10］. . 10.1017/cbo9780511761942.003
7	KIPF T N， WELLING M. Semi-supervised classification with graph convolutional networks ［EB/OL］. ［2022-06-02］. . 10.48550/arXiv.1609.02907
8	LI Y， ZEMEL R， BROCKSCHMIDT M， et al. Gated graph sequence neural networks ［EB/OL］. ［2022-03-26］. .
9	VELIKOVI P， CUCURULL G， CASANOVA A， et al. Graph attention networks ［EB/OL］. ［2022-03-28］. .
10	LEE J B， ROSSI R， KONG X. Graph classification using structural attention ［C］// Proceedings of the 24th International Conference on Knowledge Discovery and Data Mining. New York： ACM， 2018： 1666-1674. 10.1145/3219819.3219980
11	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. 10.3778/j.issn.1002-8331.2110-0345
	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. 10.3778/j.issn.1002-8331.2110-0345
12	SHANI G， HECKERMAN D， BRAFMAN R I， et al. An MDP-based recommender system ［J］. Journal of Machine Learning Research， 2005， 6： 1265-1295.
13	RENDLE S， FREUDENTHALER C， SCHMIDT-THIEME L. Factorizing personalized Markov chains for next-basket recommendation［C］// Proceedings of the 19th International Conference on World Wide Web. New York： ACM， 2010： 811-820. 10.1145/1772690.1772773
14	DAVIDSON J， LIEBALD B， LIU J， et al. The YouTube video recommendation system［C］// Proceedings of the 2010 4th ACM Conference on Recommender Systems. New York： ACM， 2010： 293-296. 10.1145/1864708.1864770
15	HIDASI B， KARATZOGLOU A， BALTRUNAS L， et al. Session-based recommendations with recurrent neural networks［EB/OL］. ［2022-04-15］. .
16	TAN Y K， XU X， LIU Y. Improved recurrent neural networks for session-based recommendations［C］// Proceedings of the 1st Workshop on Deep Learning for Recommender Systems. New York： ACM， 2016： 17-22. 10.1145/2988450.2988452
17	LI J， REN P， CHEN Z， et al. Neural attentive session-based recommendation ［C］// Proceedings of the 2017 ACM on Conference on Information and Knowledge Management. New York： ACM， 2017： 1419-1428. 10.1145/3132847.3132926
18	LIU Q， ZENG Y， MOKHOSI R， et al. STAMP： short-term attention/memory priority model for session-based recommendation ［C］// Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. New York： ACM， 2018： 1831-1839. 10.1145/3219819.3219950
19	WU S， TANG Y， ZHU Y， et al. Session-based recommendation with graph neural networks ［EB/OL］. ［2022-11-01］.. 10.1609/aaai.v33i01.3301346
20	SONG W， XIAO Z， WANG Y， et al. Session-based social recommendation via dynamic graph attention networks［C］// Proceedings of the 12th ACM International Conference on Web Search and Data Mining. New York： ACM， 2019： 555-563. 10.1145/3289600.3290989
21	YU F， ZHU Y， LIU Q， et al. TAGNN： target attentive graph neural networks for session-based recommendation［C］// Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval. New York： ACM， 2020： 1921-1924. 10.1145/3397271.3401319
22	XU C， ZHAO P， LIU Y， et al. Graph contextualized self-attention network for session-based recommendation ［C］// Proceedings of the 28th International Joint Conference on Artificial Intelligence. PaloAlto， CA： AAAI Press， 2019： 3940-3946. 10.24963/ijcai.2019/547
23	XIAN X， FANG L， SUN S. ReGNN： a repeat aware graph neural network for session-based recommendations［J］. IEEE Access， 2020， 8： 98518-98525. 10.1109/ACCESS.2020.2997722
24	CHEN T， WONG R C W. Handling information loss of graph neural networks for session-based recommendation［C］// Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. New York： ACM， 2020： 1172-1180. 10.1145/3394486.3403170
25	FANG J. Session-based Recommendation with self-attention networks［EB/OL］. ［2022-04-28］. .
26	南宁，杨程屹，武志昊. 基于多图神经网络的会话感知推荐模型［J］. 计算机应用， 2021， 41（2）： 330-336. 10.11772/j.issn.1001-9081.2020060805
	NAN N， YANG C Y， WU Z H. Multi-graph neural network-based session perception recommendation model［J］. Journal of Computer Applications， 2021， 41（2）： 330-336. 10.11772/j.issn.1001-9081.2020060805
27	任俊伟，曾诚，肖丝雨，等. 基于会话的多粒度图神经网络推荐模型［J］. 计算机应用， 2021， 41（11）： 3164-3170. 10.11772/j.issn.1001-9081.2021010060
	REN J W， ZENG C， XIAO S Y， et al. Session-based recommendation model of multi-granular graph neural network［J］. Journal of Computer Applications， 2021， 41（11）： 3164-3170. 10.11772/j.issn.1001-9081.2021010060
28	黄震华，林小龙，孙圣力，等. 会话场景下基于特征增强的图神经推荐方法［J］. 计算机学报， 2022， 45（4）： 766-780. 10.11897/SP.J.1016.2022.00766
	HUANG Z H， LIN X L， SUN S L， et al. Feature augmentation based graph neural recommendation method in session scenarios［J］. Chinese Journal of Computers， 2022， 45（4）： 766-780. 10.11897/SP.J.1016.2022.00766
29	闫昭，项欣光，李泽超.基于交互序列商品相关性建模的图卷积会话推荐［J］. 中国科学：信息科学， 2022， 52（6）： 1069-1082.
	YAN Z， XIANG X G， LI Z C. Graph convolution session recommendation based on interactive sequence commodity correlation modeling［J］. SCIENTIA SINICA Informationis， 2022， 52（6）： 1069-1082.
30	HOU Y， HU B， ZHANG Z， et al. CORE： simple and effective session-based recommendation within consistent representation space［C］// Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval. New York： ACM， 2022： 1796-1801. 10.1145/3477495.3531955
31	HUANG G， LIU Z， VAN DER MAATEN L， et al. Densely connected convolutional networks［C］// Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2017： 4700-4708. 10.1109/cvpr.2017.243

[1]	杨昊, 张轶. 基于上下文信息和多尺度融合重要性感知的特征金字塔网络算法[J]. 《计算机应用》唯一官方网站, 2023, 43(9): 2727-2734.
[2]	马国帅, 钱宇华, 张亚宇, 李俊霞, 刘郭庆. 动态异构信息融合的科研合作潜力预测[J]. 《计算机应用》唯一官方网站, 2023, 43(9): 2775-2783.
[3]	袁国龙, 张玉金, 刘洋. 基于残差反馈和自注意力的图像篡改取证网络[J]. 《计算机应用》唯一官方网站, 2023, 43(9): 2925-2931.
[4]	张秋余, 温永旺. 用于语音检索的三联体深度哈希方法[J]. 《计算机应用》唯一官方网站, 2023, 43(9): 2910-2918.
[5]	郑浩东, 马华, 谢颖超, 唐文胜. 融合遗忘因素与记忆门的图神经网络知识追踪模型[J]. 《计算机应用》唯一官方网站, 2023, 43(9): 2747-2752.
[6]	王宏, 钱清, 王欢, 龙永. 融合大核注意力卷积的轻量化图像篡改定位算法[J]. 《计算机应用》唯一官方网站, 2023, 43(9): 2692-2699.
[7]	潘润超, 虞启山, 熊泓霏, 刘智慧. 基于深度图神经网络的协同推荐算法[J]. 《计算机应用》唯一官方网站, 2023, 43(9): 2741-2746.
[8]	梁美佳, 刘昕武, 胡晓鹏. 基于改进YOLOv3的列车运行环境图像小目标检测算法[J]. 《计算机应用》唯一官方网站, 2023, 43(8): 2611-2618.
[9]	马胜位, 黄瑞章, 任丽娜, 林川. 基于多层语义融合的结构化深度文本聚类模型[J]. 《计算机应用》唯一官方网站, 2023, 43(8): 2364-2369.
[10]	王静红, 周志霞, 王辉, 李昊康. 双路自编码器的属性网络表示学习[J]. 《计算机应用》唯一官方网站, 2023, 43(8): 2338-2344.
[11]	衡红军, 杨鼎诚. 知识增强的方面词交互图神经网络[J]. 《计算机应用》唯一官方网站, 2023, 43(8): 2412-2419.
[12]	崔雨萌, 王靖亚, 刘晓文, 闫尚义, 陶知众. 融合注意力和裁剪机制的通用文本分类模型[J]. 《计算机应用》唯一官方网站, 2023, 43(8): 2396-2405.
[13]	齐爱玲, 王宣淋. 基于中层细微特征提取与多尺度特征融合细粒度图像识别[J]. 《计算机应用》唯一官方网站, 2023, 43(8): 2556-2563.
[14]	张琨, 杨丰玉, 钟发, 曾广东, 周世健. 基于混合代码表示的源代码脆弱性检测[J]. 《计算机应用》唯一官方网站, 2023, 43(8): 2517-2526.
[15]	金泽熙, 李磊, 刘继. 基于改进领域分离网络的迁移学习模型[J]. 《计算机应用》唯一官方网站, 2023, 43(8): 2382-2389.

融合项目影响力的图神经网络会话推荐模型

Session-based recommendation model by graph neural network fused with item influence

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 10

参考文献 31

相关文章 15

编辑推荐

Metrics