融合双端注意力网络的知识图谱推荐算法

doi:10.3778/j.issn.1002-8331.2206-0073

摘要/Abstract

摘要： 知识图谱在推荐算法中的应用有效增强了推荐结果的可解释性，但仍缺乏对用户-物品交互信息的关注。针对目前已有的知识图谱推荐算法忽略用户端与物品端的差异而无法有效地提取用户与物品信息的问题，提出了融合双端注意力网络的知识图谱推荐算法（double end knowledge graph attention network，DEKGAN）。首先根据用户端与物品端两者不同的需求设计两种注意力网络，在用户端是将用户的浏览记录输入到知识图谱中不断扩展用户的潜在兴趣，通过基于用户的注意力网络对三元组内部信息进行计算以更准确地获取用户的嵌入表示向量；在物品端是将物品信息输入到知识图谱内中不断获取与用户喜好有关的物品信息，使用基于用户偏好的注意力网络指定更准确的传播方向，然后将两端注意力网络中获取到的邻域信息聚合生成物品的嵌入表示向量，最终获取用户对物品的喜好概率。通过在数据集MovieLens-1M和Book-Crossing中进行实验，采用AUC、F1、precision和recall指标进行评估，结果证明与其他基准算法相比推荐的准确性和可解释性有显著的提升。

关键词: 知识图谱, 双端采样, 偏好传播, 注意力网络

Abstract: The application of knowledge graphs in recommendation algorithms effectively enhances the interpretability of recommendation results, but still lacks attention to user-item interaction information. Aiming at the problem that the existing knowledge graph recommendation algorithms ignore the difference between the user end and the item end and cannot effectively extract user and item information, a knowledge graph recommendation algorithm integrating double-end attention network （double end knowledge graph attention network, DEKGAN） is proposed. Firstly, two attention networks are designed according to the different needs of the user end and the item end. On the user side, the browsing records of user are input into the knowledge graph to continuously expand the potential interest of user, and the internal information of triples is calculated by user-based attention network to obtain the embedded representation vector of user more accurately. On the item side, the item information is input into the knowledge graph to continuously obtain the item information related to the preference of user, and the attention network based on the preference of user is used to specify a more accurate propagation direction.Then, the neighborhood information obtained in the attention networks at both ends is aggregated to generate the embedding representation vector of the item, and finally the preference of user probability for the item is obtained. By conducting experiments on the datasets MovieLens-1M and Book-Crossing, using AUC, F1, precision and recall metrics for evaluation, the results demonstrate a significant improvement in recommendation accuracy and interpretability compared to other benchmark algorithms.

Key words: knowledge graph, paired sampling, preference propagation, attention network

王光, 石山山. 融合双端注意力网络的知识图谱推荐算法[J]. 计算机工程与应用, 2023, 59(19): 114-121.

WANG Guang, SHI Shanshan. Knowledge Graph Recommendation Algorithm Integrating Double-End Attention Network[J]. Computer Engineering and Applications, 2023, 59(19): 114-121.

参考文献

[1] KAMEHKHOSH I，JANNACH D.User perception of next-track music recommendations[C]//Proceedings of the 25th Conference on User Modeling，Adaptation and Personalization，2017.
[2] CHOUDHURY S S，MOHANTY S N，JAGADEV A K. Multimodal trust based recommender system with machine learning approaches for movie recommendation[J].International Journal of Information Technology，2021，13（2）：475-482.
[3] 耿化聪，梁宏涛，刘国柱.基于知识图谱与协同过滤的饮食推荐算法[J].计算机与现代化，2021（8）：24-29.
GENG H C，LIANG H T，LIU G Z.Diet recommendation algorithm based on knowledge graph and collaborative filtering[J].Computer and Modernization，2021（8）：24-29.
[4] RODRIGUEZ-PARDO C，PATRICIO M A，BERLANGA A，et al.Machinelearning for smart tourism and rctail[M]//Handbook of Researchon Eig Data Clustering and Machine Learning.Hershey：1G1Global，2020：311-333.
[5] YAO L，XU Z，ZHOU X，et al. Synergies between association rules and collaborative filtering in recommender system：an application to auto industry[J].Data Science and Digital Business，2019，2560：65-80.
[6] HUANG Y，HUANG W J，XIANG X L，et al.An empirical study of personalized advertising recommendation based on DBSCAN clustering of sina weibo user-generated content[J].Procedia Computer Science，2021，183（8）：303-310.
[7] HUANG W，SU X，WU M，et al.Category，process，and recommendation of design in an interactive evolutionary computation interior design experiment：a data-driven study[J].Artificial Intelligence for Engineering Design Analysis and Manufacturing，2020，34（2）：1-15.
[8] SANG L，XU M，QIAN S，et al.Knowledge graph enhanced neural collaborative recommendation[J].Expert Systems with Applications，2021，164（12）：113992.
[9] YU B，ZHOU C，ZHANG C，et al.A privacy-preserving multi-task framework for knowledge graph enhanced recommendation[J].IEEE Access，2020，8：115717-115727.
[10] WANG H W，ZHANG F Z，XIE X，et al.DKN：deep knowledge-aware network for news recommendation[C]//Proceedings of the 2018 World Wide Web Conference，Lyon，France，2018：1835-1844.
[11] CAO Y，WANG X，HE X，ET al.Unifying knowledge graph learning and recommendation：towards a better understanding of user preferences[C]//Proceedings of the World Wide Web Conference，2019：151-161.
[12] SHA X，SUN Z，ZHANG J.Hierarchical attentive knowledge graph embedding for personalized recommendation[J].Electronic Commerce Research and Applications，2021，48：101071.
[13] SUN Z，YANG J，ZHANG J，et al.Recurrent knowledge graph embedding for effective recommendation[C]//Proceedings of the 12th ACM Conference on Recommender Systems，2018：297-305.
[14] HUANG X，FANG Q，QIAN S，et al.Explainable interaction-driven user modeling over knowledge graph for sequential recommendation[C]//Proceedings of the the 27th ACM International Conference，2019：548-556.
[15] WANG H，ZHANG F，WANG J，et al.RippleNet：propagating user preferences on the knowledge graph for recommender systems[C]//Proceedings of the 27th ACM International Conference on Information and Knowledge Management，Torino，Oct 22-26，2018.New York：ACM，2018：417-426.
[16] WANG H，ZHAO M，XIE X，et al.Knowledge graph convolutional networks for recommender systems[C]//Proceedings of the World Wide Web Conference，San Francisco，May 13?17，2019.New York：ACM，2019：3307-3313.
[17] WANG X，HE X，CAO Y，et al.KGAT：knowledge graph attention network for recommendation[C]//Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining，Anchorage，Aug 4-8，2019.New York：ACM，2019： 950-958.
[18] 樊海玮，张锐驰，安毅生，等.融合知识图谱邻居双端的在线学习资源推荐算法[J].计算机应用，2022（4）：1-6.
FAN H W，ZHANG R C，AN Y S，et al.Online learning resource recommendation algorithm integrating both sides of knowledge graph neighborhood[J].Journal of Computer Applications，2022（4）：1-6
[19] 李想，杨兴耀，于炯，等.基于知识图谱卷积网络的双端推荐算法[J].计算机科学与探索，2022，16（1）：176-184.
LI X，YANG X Y，YU J，et al.Double end knowledge graph convolutional networks for recommender systems[J].Journal of Frontiers of Computer Science and Technology，2022，16（1）：176-184.
[20] 许杰，祝玉坤，邢春晓.基于深度强化学习的金融交易算法研究[J].计算机工程与应用，2022，58（7）：276-285.
XU J，ZHU Y K，XING C X.Research on financial trading algorithm based on deep reinforcement learning[J].Computer Engineering and Applications，2022，58（7）：276-285.
[21] 阎世宏，马为之，张敏，等.结合用户长短期兴趣的深度强化学习推荐方法[J].中文信息学报，2021，35（8）：107-116.
YAN S H，MA W Z，ZHANG M，et al.Deep reinforcement learning recommendation method combined with users’ long-term and short-term interests[J].Chinese Journal of Information，2021，35（8）：107-116.
[22] 史存会，胡耀康，冯彬，等.舆情场景下基于层次知识的话题推荐方法[J].计算机研究与发展，2021，58（8）：1811-1819.
SHI C H，HU Y K，FENG B，et al.A hierarchical knowledge based topic recommendation method in public opinion scenario[J].Journal of Computer Research and Development，2021，58（8）：1811-1819.
[23] WANG Z，LIN G Y，TAN H B，et al.CKAN：collaborative knowledge-aware attentive network for recommender systems[C]//Proceedings of the 43st International ACM SIGIR Conference on Research & Development in Information Retrieval.New York：ACM，2020：219-228.
[24] WANG H，ZHANG F，ZHANG M，et al.Knowledge aware graph neural networks with label smoothness regularization for recommender systems[C]//Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining，Anchorage Aug 4-8，2019. New York：ACM，2019：968-977.
[25] TU K，CUI P，WANG D X，et al.Conditional attention networks for distilling knowledge graphs in recommendation[J].arXiv：2111.02100，2021.