TY - GEN
T1 - Knowledge Graph Attention Network Enhanced Sequential Recommendation
AU - Zhu, Xingwei
AU - Zhao, Pengpeng
AU - Xu, Jiajie
AU - Fang, Junhua
AU - Zhao, Lei
AU - Xian, Xuefeng
AU - Cui, Zhiming
AU - Sheng, Victor S.
N1 - Publisher Copyright:
© 2020, Springer Nature Switzerland AG.
PY - 2020
Y1 - 2020
N2 - Knowledge graph (KG) has recently been proved effective and attracted a lot of attentions in sequential recommender systems. However, the relations between the attributes of different entities in KG, which could be utilized to improve the performance, remain largely unexploited. In this paper, we propose an end-to-end Knowledge Graph attention network enhanced Sequential Recommendation (KGSR) framework to capture the context-dependency of sequence items and the semantic information of items in KG by explicitly exploiting high-order relations between entities. Specifically, our method first combines the user-item bipartite graph and the KG into a unified graph and encodes all nodes of the unified graph into vector representations with TransR. Then, a graph attention network recursively propagates the information of neighbor nodes to refine the embedding of nodes and distinguishes the importance of neighbors with an attention mechanism. Finally, we apply recurrent neural network to capture the user’s dynamic preferences by encoding user-interactive sequence items that contain rich auxiliary semantic information. Experimental results on two datasets demonstrate that KGSR outperforms the state-of-the-art sequential recommendation methods.
AB - Knowledge graph (KG) has recently been proved effective and attracted a lot of attentions in sequential recommender systems. However, the relations between the attributes of different entities in KG, which could be utilized to improve the performance, remain largely unexploited. In this paper, we propose an end-to-end Knowledge Graph attention network enhanced Sequential Recommendation (KGSR) framework to capture the context-dependency of sequence items and the semantic information of items in KG by explicitly exploiting high-order relations between entities. Specifically, our method first combines the user-item bipartite graph and the KG into a unified graph and encodes all nodes of the unified graph into vector representations with TransR. Then, a graph attention network recursively propagates the information of neighbor nodes to refine the embedding of nodes and distinguishes the importance of neighbors with an attention mechanism. Finally, we apply recurrent neural network to capture the user’s dynamic preferences by encoding user-interactive sequence items that contain rich auxiliary semantic information. Experimental results on two datasets demonstrate that KGSR outperforms the state-of-the-art sequential recommendation methods.
KW - Graph neural network
KW - Knowledge graph
KW - Sequential recommendation
UR - http://www.scopus.com/inward/record.url?scp=85093960564&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-60259-8_15
DO - 10.1007/978-3-030-60259-8_15
M3 - Conference contribution
AN - SCOPUS:85093960564
SN - 9783030602581
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 181
EP - 195
BT - Web and Big Data - 4th International Joint Conference, APWeb-WAIM 2020, Proceedings
A2 - Wang, Xin
A2 - Zhang, Rui
A2 - Lee, Young-Koo
A2 - Sun, Le
A2 - Moon, Yang-Sae
PB - Springer Science and Business Media Deutschland GmbH
T2 - 4th Asia-Pacific Web and Web-Age Information Management, Joint Conference on Web and Big Data, APWeb-WAIM 2020
Y2 - 18 September 2020 through 20 September 2020
ER -