Graph neural networks can fuse node information and topology, and are widely used in recommendation algorithms in recent years. However, the existing recommendation models based on graph neural networks have coarse granularity when modeling user behavior, and the user feature learning algorithm lacks consideration of historical information, both of which hinder the extraction of user preference features. To address the above problems, this paper proposes an enhanced multi-channel graph attention based collaborative filtering recommendation model(EMGACF). In the neighborhood aggregation phase, multi-channel graph attention is used to model fine-grained user rating levels, which effectively improves the learning ability of the model for user preferences; in the node update phase, an - DOI:10.3969/j.issn.1000-1565.2024.02.010基于增强多通道图注意力的推荐模型张昱1,2,苏仡琳1,李继涛1,陈广书1,张明魁1(1.北京建筑大学 电气与信息工程学院,建筑大数据智能处理方法研究北京市重点实验室,北京 100044;2.中国矿业大学 深地空间科学与工程研究院,深部岩土力学与地下工程国家重点实验室,北京 100083)摘 要:图神经网络具备融合节点信息与拓扑结构的能力,近年来在推荐算法中得到了广泛的应用.然而,现有的基于图神经网络的推荐模型用户行为建模粒度较粗,用户特征学习算法对历史信息使用不足,两者阻碍了用户偏好特征的提取.针对以上问题,本文提出一种基于增强多通道图注意力的推荐模型(enhanced multi-channel graph attention based collaborative filtering recommendation model,EMGACF).在邻域聚合部分,采用多通道图注意力对细粒度用户评分等级建模,有效提升了模型对用户偏好的学习能力;在节点更新部分,提出基于增强自信息的节点更新算法,使用邻居节点聚合表示的同时保留了节点自身历史信息和内在偏好,提升了迭代过程中用户偏好的学习效果.实验部分在4种规模的常用推荐系统基准数据集上训练模型,实验结果表明,预测误差相比于主流模型降低了1.43%~7.81%.关键词:图注意力;用户偏好;自编码器;协同过滤中图分类号:TP391 文献标志码:A 文章编号:1000-1565(2024)02-0190-09Enhanced multi-channel graph attention based on recommendation modelZHANG Yu1,2, SU Yilin1, LI Jitao1, CHEN Guangshu1, ZHANG Mingkui1(1. Beijing Key Laboratory of Intelligent Processing for Building Big Data, School of Electrical and Information Engineering, Beijing University of Civil Engineering and Architecture, Beijing 100044, China;2. State Key Laboratory in China for GeoMechanics and Deep Underground Engineering, Institute of Deep Underground Space Science and Engineering, China University of Mining and Technology, Beijing 100083, China)Abstract: Graph neural networks can fuse node information and topology, and are widely used in recommendation algorithms in recent years. However, the existing recommendation models based on graph neural networks have coarse granularity when modeling user behavior, and the user feature learning algorithm lacks consideration of historical information, both of which hinder the extraction of user preference features. To address the above problems, this paper proposes an enhanced multi-channel graph attention based collaborative filtering recommendation model(EMGACF). In the neighborhood aggregation phase, multi-channel graph attention is used to model fine-grained user rating levels, which effectively improves the learning ability of the model for user preferences; in the node update phase, an - 收稿日期:2023-10-31;修回日期:2024-01-13 基金项目:国家重点实验室深地空间科学与工程研究院基金资助项目(XD2021021);北京建筑大学2022年度研究生教育教学质量提升资助项目(J2022003) 第一作者:张昱(1979—),男,北京建筑大学副教授,博士,主要从事大数据、人工智能、智慧城市与岩爆研究.E-mail: bigdata@bucea.edu.cn 通信作者:苏仡琳(1997—),女,北京建筑大学在读硕士研究生,主要从事大数据和人工智能、推荐算法研究.E-mail: syl9790@163.com第2期张昱等:基于增强多通道图注意力的推荐模型enhanced self-information-based node update algorithm is proposed, which uses the aggregated representation of neighboring nodes while preserving the nodes' own historical information and intrinsic preferences. This improves the learning effect of user preferences in the iterative process. By training the model on three benchmark datasets of different sizes of recommender systems, the experimental results show that the prediction error is reduced by 1.43% to 7.81% compared with the mainstream model.
ZHANG Yu
,
SU Yilin
,
LI Jitao
,
CHEN Guangshu
,
ZHANG Mingkui
. Enhanced multi-channel graph attention based on recommendation model[J]. Journal of Hebei University(Natural Science Edition), 2024
, 44(2)
: 190
-198
.
DOI: 10.3969/j.issn.1000-1565.2024.02.010
[1] ADOMAVICIUS G, TUZHILIN A. Toward the next generation of recommender systems: a survey of the state-of-the-art and possible extensions[J]. IEEE Trans Knowl Data Eng, 2005, 17(6): 734-749. DOI: 10.1109/TKDE.2005.99.
[2] VERBERT K, MANOUSELIS N, OCHOA X, et al. Context-aware recommender systems for learning: a survey and future challenges[J]. IEEE Trans Learn Technol, 2012, 5(4): 318-335. DOI: 10.1109/TLT.2012.11.
[3] MOONEY R J, ROY L. Content-based book recommending using learning for text categorization[C] //Proceedings of the fifth ACM conference on Digital libraries, June 2-7, 2000, San Antonio, Texas, USA. ACM, 2000: 195-204. DOI: 10.1145/336597.336662.
[4] BREESE J S, HECKERMAN D, KADIE C. Empirical analysis of predictive algorithms for collaborative filtering[J]. ArXiv Preprint ArXiv:1301.7363, 2013.DOI:10.48550/arXiv.1301.7363.
[5] BALABANOVIC' M, SHOHAM Y. Fab[J]. Commun ACM, 1997, 40(3): 66-72. DOI: 10.1145/245108.245124.
[6] 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. DOI: 10.11897/SP.J.1016.2018.01619.
[7] LECUN Y, BENGIO Y, HINTON G. Deep learning[J]. Nature, 2015, 521: 436-444. DOI: 10.1038/nature14539.
[8] WU S W, SUN F, ZHANG W T, et al. Graph neural networks in recommender systems: a survey[J]. ACM Comput Surv, 2022, 55(5): 97. DOI: 10.1145/3535101.
[9] 林晓,周云翔,李大志,等.利用多尺度特征联合注意力模型的图像修复[J].计算机辅助设计与图形学学报, 2022, 34(8): 1260-1271. DOI: 10.3724/SP.J.1089.2022.19172.
[10] 柴玉梅,员武莲,王黎明,等.基于双注意力机制和迁移学习的跨领域推荐模型[J].计算机学报, 2020, 43(10): 1924-1942. DOI: 10.11897/SP.J.1016.2020.01924.
[11] KIPF T N, WELLING M. Variational graph auto-encoders[J]. ArXiv Preprint ArXiv:1611.07308,2016.DOI:10.48550/arXiv.1611.07308.
[12] SALHA G, HENNEQUIN R, VAZIRGIANNIS M. Keep it simple: Graph autoencoders without graph convolutional networks[J]. ArXiv Preprint ArXiv:1910.00942, 2019.DOI:10.48550/arXiv.1910.00942.
[13] LI I, FABBRI A, HINGMIRE S, et al. R-vgae: Relational-variational graph autoencoder for unsupervised prerequisite chain learning[J]. ArXiv Preprint ArXiv:2004.10610, 2020.DOI: 10.48550/arXiv.2004.10610.
[14] VELI(ˇoverC)KOVIC' P,CUCURULL G,CASANOVA A,et al.Graph attention networks[J]. ArXiv Preprint ArXiv:1710.10903,2017.DOI: 10.48550/arXiv.1710.10903.
[15] MA C, KANG P, WU B, et al. Gated attentive-autoencoder for content-aware recommendation[C] //Proceedings of the Twelfth ACM International Conference on Web Search and Data Mining, February 11 - 15, 2019, Melbourne VIC, Australia. ACM, 2019: 519-527. DOI: 10.1145/3289600.3290977.
[16] YU J L, YIN H Z, LI J D, et al. Enhancing social recommendation with adversarial graph convolutional networks[J]. IEEE Trans Knowl Data Eng, 2022, 34(8): 3727-3739. DOI: 10.1109/TKDE.2020.3033673.
[17] XU J X, ZHU Z L, ZHAO J X, et al. Gemini: a novel and universal heterogeneous graph information fusing framework for online recommendations[C] //Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, July 6 - 10, 2020, Virtual Event, CA, USA. ACM, 2020: 3356-3365. DOI: 10.1145/3394486.3403388.
[18] WANG X, WANG R J, SHI C, et al. Multi-component graph convolutional collaborative filtering[J]. Proc AAAI Conf Artif Intell, 2020, 34(4): 6267-6274. DOI: 10.1609/aaai.v34i04.6094.
[19] BERG R, KIPF T N, WELLING M. Graph convolutional matrix completion[J]. ArXiv Preprint ArXiv:1706.02263,2017.DOI: 10.48550/arXiv.1706.02263.
[20] CANDÈS E J, RECHT B. Exact matrix completion via convex optimization[J]. Found Comput Math, 2009, 9(6): 717-772. DOI: 10.1007/s10208-009-9045-5.
[21] MONTI F, BRONSTEIN M M, BRESSON X. Geometric matrix completion with recurrent multi-graph neural networks[C] //Proceedings of the 31st International Conference on Neural Information Processing Systems, December 4-9, 2017, Long Beach, California, USA. ACM, 2017: 3700-3710. DOI: 10.5555/3294996.3295127.
[22] ZHENG Y, TANG B S, DING W K, et al. A neural autoregressive approach to collaborative filtering[C] //Proceedings of the 33rd International Conference on International Conference on Machine Learning - Volume 48, June 19 - 24, 2016, New York, NY, USA. ACM, 2016: 764-773. DOI: 10.5555/3045390.3045472.
[23] LI Q, ZHENG X, WU X. Collaborative autoencoder for recommender systems[J]. ArXiv E-prints, 2017.DOI:10.1145/3097983.3098077. (
