基于解耦图神经网络的可解释标签感知推荐算法
作者:
作者简介:

杜晓宇(1986-),男,博士,副教授,CCF专业会员,主要研究领域为多媒体推荐,机器学习.;陈正(1995-),男,硕士,主要研究领域为推荐系统.;项欣光(1982-),男,博士,副教授,CCF专业会员,主要研究领域为图像处理,计算机视觉,多媒体技术.

通讯作者:

项欣光,E-mail:xgxiang@njust.edu.cn

中图分类号:

TP18

基金项目:

国家自然科学基金(62172226); 江苏省“双创博士”人才计划 (JSSCBS20210200)


Explainable Tag-aware Recommendation Based on Disentangled Graph Neural Network
Author:
  • 摘要
  • | |
  • 访问统计
  • |
  • 参考文献 [37]
  • |
  • 相似文献 [20]
  • |
  • 引证文献
  • | |
  • 文章评论
    摘要:

    标签感知推荐算法利用标签标注数据提升推荐模型对用户偏好和项目属性的理解, 受到业界的广泛关注. 但是, 现有方法常忽视了用户关注点、项目属性和标签含义的多样性, 干扰了三者关系推断, 从而影响推荐结果. 因此, 提出一种基于解耦图神经网络的可解释标签感知推荐算法(DETRec), 解构用户、项目和标签的关注角度, 并由此形成可解释的推荐依据. 具体来讲, DETRec构造关系图以建模用户、项目和标签的关系; 通过邻域路由机制和消息传播机制, 分离结点形成属性子图, 以描述不同属性下的结点关系; 最终根据属性子图形成推荐依据. 实现了两种DETRec实例: 单图实例(DETRec-S)在单个关系图中描述全部结点关系; 多图实例(DETRec-M)使用3个二分图分别描述用户-项目、项目-标签、用户-标签关系. 在3个公开数据集上进行的大量实验表明, DETRec的两种实例均明显优于标签感知推荐的基准模型, 也为推荐结果生成了对应的推荐依据, 是有效的可解释标签感知推荐算法.

    Abstract:

    Tag-aware recommendation algorithms use tagged data to enhance the recommendation models’ understanding of user preferences and item attributes, which attract extensive attention in the field. Most existing methods, however, neglect the diversities of user concerns, item attributes, and tag semantics and interfere with the correlation inference of the three, which affects the recommendation results. Therefore, this study introduces the disentangled graph neural network (DGNN) method into the tag-aware recommendation task and proposes a DGNN-based explainable tag-aware recommendation (DETRec) method. It disentangles the perspectives of users, items, and tags to provide explainable recommendation references. Specifically, DETRec utilizes a correlation graph construction module to model the user-item-tag correlations. Then, it employs a neighborhood routing mechanism and a message propagation mechanism to disentangle the nodes to form the sub-graphs of attributes and thereby describe the nodal correlations under different attributes. Finally, it generates recommendation references on the basis of these attribute sub-graphs. This study implements two types of DETRec instantiation: 1) DETRec based on a single graph (DETRec-S), which describes all correlations of user, item, and tag nodes in a single graph, and 2) DETRec based on multiple graphs (DETRec-M), which utilizes three bipartite graphs to describe the user-item, item-tag, and user-tag correlations separately. Extensive experiments on three public datasets demonstrate that the above two types of DETRec instantiation are significantly superior to the baseline model and generate the references corresponding to the recommendation results. Hence, DETRec is an effective explainable tag-aware recommendation algorithm.

    参考文献
    [1] 邢千里, 刘列, 刘奕群, 张敏, 马少平. 微博中用户标签的研究. 软件学报, 2015, 26(7): 1626-1637. http://www.jos.org.cn/1000-9825/4655.htm
    Xing QL, Liu L, Liu YQ, Zhang M, Ma SP. Study on user tags in Weibo. Ruan Jian Xue Bao/Journal of Software, 2015, 26(7): 1626-1637 (in Chinese with English abstract). http://www.jos.org.cn/1000-9825/4655.htm
    [2] Shoja BM, Tabrizi N. Tags-aware recommender systems: A systematic review. In: Proc. of the IEEE Int’l Conf. on Big Data, Cloud Computing, Data Science & Engineering (BCD). Honolulu: IEEE, 2019. 11–18.
    [3] Su XY, Khoshgoftaar TM. A survey of collaborative filtering techniques. Advances in Artificial Intelligence, 2009, 2009: 421425. [doi: 10.1155/2009/421425]
    [4] Wu L, He XN, Wang X, Zhang K, Wang M. A survey on accuracy-oriented neural recommendation: From collaborative filtering to information-rich recommendation. IEEE Trans. on Knowledge and Data Engineering, 2022: 1–20.
    [5] Zuo Y, Zeng JL, Gong MG, Jiao LC. Tag-aware recommender systems based on deep neural networks. Neurocomputing, 2016, 204: 51–60. [doi: 10.1016/j.neucom.2015.10.134]
    [6] Xu ZH, Chen C, Lukasiewicz T, Miao YS, Meng XW. Tag-aware personalized recommendation using a deep-semantic similarity model with negative sampling. In: Proc. of the 25th ACM Int’l on Conf. on Information and Knowledge Management. Indianapolis: ACM, 2016. 1921–1924.
    [7] Chen B, Guo W, Tang RM, Xin X, Ding Y, He XQ, Wang D. TGCN: Tag graph convolutional network for tag-aware recommendation. In: Proc. of the 29th ACM Int’l Conf. on Information & Knowledge Management. ACM, 2020. 155–164.
    [8] 张斌, 张引, 高克宁, 郭朋伟, 孙达明. 融合关系与内容分析的社会标签推荐. 软件学报, 2012, 23(3): 476-488. http://www.jos.org.cn/1000-9825/4001.htm
    Zhang B, Zhang Y, Gao KN, Guo PW, Sun DM. Combining relation and content analysis for social tagging recommendation. Ruan Jian Xue Bao/Journal of Software, 2012, 23(3): 476-488 (in Chinese with English abstract). http://www.jos.org.cn/1000-9825/4001.htm
    [9] 徐鹏宇, 刘华锋, 刘冰, 景丽萍, 于剑. 标签推荐方法研究综述. 软件学报, 2022, 33(4): 1244–1266. http://www.jos.org.cn/1000-9825/6481.htm
    Xu PY, Liu HF, Liu B, Jing LP, Yu J. Survey of tag recommendation methods. Ruan Jian Xue Bao/Journal of Software, 2022, 33(4): 1244–1266 (in Chinese with English abstract). http://www.jos.org.cn/1000-9825/6481.htm
    [10] Xu ZH, Lukasiewicz T, Chen C, Miao YS, Meng XW. Tag-aware personalized recommendation using a hybrid deep model. In: Proc. of the 26th Int’l Joint Conf. on Artificial Intelligence. Melbourne: AAAI Press, 2017. 3196–3202.
    [11] Huang RR, Wang N, Han CQ, Yu F, Cui L. TNAM: A tag-aware neural attention model for Top-N recommendation. Neurocomputing, 2020, 385: 1–12. [doi: 10.1016/j.neucom.2019.11.095]
    [12] 王瑜, 武延军, 吴敬征, 刘晓燕. 基于异构网络面向多标签系统的推荐模型研究. 软件学报, 2017, 28(10): 2611–2624. http://www.jos.org.cn/1000-9825/5132.htm
    Wang Y, Wu YJ, Wu JZ, Liu XY. Multi-Dimensional tag recommender model via heterogeneous networks. Ruan Jian Xue Bao/Journal of Software, 2017, 28(10): 2611–2624 (in Chinese with English abstract). http://www.jos.org.cn/1000-9825/5132.htm
    [13] 马慧芳, 张迪, 赵卫中, 史忠植. 基于超图随机游走标签扩充的微博推荐方法. 软件学报, 2019, 30(11): 3397-3412. http://www.jos.org.cn/1000-9825/5545.htm
    Ma HF, Zhang D, Zhao WZ, Shi ZZ. Microblog recommendation method based on hypergraph random walk tag extension. Ruan Jian Xue Bao/Journal of Software, 2019, 30(11): 3397-3412 (in Chinese with English abstract). http://www.jos.org.cn/1000-9825/5545.htm
    [14] Du XY, He XN, Yuan FJ, Tang JH, Qin ZG, Chua TS. Modeling embedding dimension correlations via convolutional neural collaborative filtering. ACM Transactions on Information Systems, 2019, 37(4): 47. [doi: 10.1145/3357154]
    [15] Du XY, Wu ZK, Feng FL, He XN, Tang JH. Invariant representation learning for multimedia recommendation. In: Proc. of the 30th ACM Int’l Conf. on Multimedia. Lisboa: ACM, 2022. 619–628.
    [16] 葛尧, 陈松灿. 面向推荐系统的图卷积网络. 软件学报, 2020, 31(4): 1101–1112. http://www.jos.org.cn/1000-9825/5928.htm
    Ge Y, Chen SC. Graph convolutional network for recommender systems. Ruan Jian Xue Bao/Journal of Software, 2020, 31(4): 1101–1112 (in Chinese with English abstract). http://www.jos.org.cn/1000-9825/5928.htm
    [17] Wu SW, Sun F, Zhang WT, Xie X, Cui B. Graph neural networks in recommender systems: A survey. ACM Computing Surveys, 2023, 55(5): 1–37.
    [18] Wang X, He XN, Wang M, Feng FL, Chua TS. Neural graph collaborative filtering. In: Proc. of the 42nd Int’l ACM SIGIR Conf. on Research and Development in Information Retrieval. Paris: ACM, 2019. 165–174.
    [19] He XN, Deng K, Wang X, Li Y, Zhang YD, Wang M. LightGCN: Simplifying and powering graph convolution network for recommendation. In: Proc. of the 43rd Int’l ACM SIGIR Conf. on Research and Development in Information Retrieval. ACM, 2020. 639–648.
    [20] Chen M, Wei ZW, Huang ZF, Ding BL, Li YL. Simple and deep graph convolutional networks. In: Proc. of the 37th Int’l Conf. on Machine Learning. PMLR, 2020, 119: 1725–1735.
    [21] Wu F, Souza AH Jr, Zhang TY, Fifty C, Yu T, Weinberger KQ. Simplifying graph convolutional networks. In: Proc. of the 36th Int’l Conf. on Machine Learning. Long Beach: PMLR, 2019. 6861–6871.
    [22] Wei YW, Wang X, Nie LQ, He XN, Hong RC, Chua TS. MMGCN: Multi-modal graph convolution network for personalized recommendation of micro-video. In: Proc. of the 27th ACM Int’l Conf. on Multimedia. Nice: ACM, 2019. 1437–1445.
    [23] Wang X, He XN, Cao YX, Liu M, Chua TS. KGAT: Knowledge graph attention network for recommendation. In: Proc. of the 25th ACM SIGKDD Int’l Conf. on Knowledge Discovery & Data Mining. Anchorage: ACM, 2019. 950–958.
    [24] Zhao ZY, Zhang XJ, Zhou H, Li C, Gong MG, Wang YQ. HetNERec: Heterogeneous network embedding based recommendation. Knowledge-Based Systems, 2020, 204: 106218. [doi: 10.1016/j.knosys.2020.106218]
    [25] Ma JX, Cui P, Kuang K, Wang X, Zhu WW. Disentangled graph convolutional networks. In: Proc. of the 36th Int’l Conf on Machine Learning. Long Beach: PMLR, 2019. 4212–4221.
    [26] Wang X, Jin HY, Zhang A, He XN, Xu T, Chua TS. Disentangled graph collaborative filtering. In: Proc. of the 43rd Int’l ACM SIGIR Conf. on Research and Development in Information Retrieval. ACM, 2020. 1001–1010.
    [27] Wang YF, Tang SY, Lei YT, Song WP, Wang S, Zhang M. DisenHAN: Disentangled heterogeneous graph attention network for recommendation. In: Proc. of the 29th ACM Int’l Conf. on Information & Knowledge Management. ACM, 2020. 1605–1614.
    [28] Rendle S, Freudenthaler C, Gantner Z, Schmidt-Thieme L. BPR: Bayesian personalized ranking from implicit feedback. In: Proc. of the 25th Conf. on Uncertainty in Artificial Intelligence. Montreal: AUAI Press, 2009. 452–461.
    [29] Du XY, Wang X, He XN, Li ZC, Tang JH, Chua TS. How to learn item representation for cold-start multimedia recommendation? In: Proc. of the 28th ACM Int’l Conf. on Multimedia. Seattle: ACM, 2020. 3469–3477.
    [30] Cantador I, Brusilovsky P, Kuflik T. Second workshop on information heterogeneity and fusion in recommender systems (HetRec2011). In: Proc. of the 5th ACM Conf. on Recommender systems. Chicago: ACM, 2011. 387–388.
    [31] Alon U, Yahav E. On the bottleneck of graph neural networks and its practical implications. In: Proc. of the 9th Int’l Conf. on Learning Representations. OpenReview.net, 2021. 1–16.
    引证文献
    网友评论
    网友评论
    分享到微博
    发 布
引用本文

杜晓宇,陈正,项欣光.基于解耦图神经网络的可解释标签感知推荐算法.软件学报,2023,34(12):5670-5685

复制
分享
文章指标
  • 点击次数:911
  • 下载次数: 2875
  • HTML阅读次数: 1134
  • 引用次数: 0
历史
  • 收稿日期:2022-01-21
  • 最后修改日期:2022-06-11
  • 在线发布日期: 2023-02-15
  • 出版日期: 2023-12-06
文章二维码
您是第19985195位访问者
版权所有:中国科学院软件研究所 京ICP备05046678号-3
地址:北京市海淀区中关村南四街4号,邮政编码:100190
电话:010-62562563 传真:010-62562533 Email:jos@iscas.ac.cn
技术支持:北京勤云科技发展有限公司

京公网安备 11040202500063号