Recurrent knowledge graph embedding for effective recommendation

Recurrent knowledge graph embedding for effective recommendation

Sun, Zhu (Nanyang Technological University)
Yang, J. (University of Fribourg)
Zhang, J. (Nanyang Technological University)
Bozzon, A. (TU Delft Web Information Systems)
Huang, Long Kai (Nanyang Technological University)
Xu, Chi (Singapore Institute of Manufacturing Technology)

2018

Knowledge graphs (KGs) have proven to be effective to improve recommendation. Existing methods mainly rely on hand-engineered features from KGs (e.g., meta paths), which requires domain knowledge. This paper presents RKGE, a KG embedding approach that automatically learns semantic representations of both entities and paths between entities for characterizing user preferences towards items. Specifically, RKGE employs a novel recurrent network architecture that contains a batch of recurrent networks to model the semantics of paths linking a same entity pair, which are seamlessly fused into recommendation. It further employs a pooling operator to discriminate the saliency of different paths in characterizing user preferences towards items. Extensive validation on real-world datasets shows the superiority of RKGE against state-of-the-art methods. Furthermore, we show that RKGE provides meaningful explanations for recommendation results.

Attention Mechanism
Knowledge Graph
Recurrent Neural Network
Semantic Representation

http://resolver.tudelft.nl/uuid:9a3559e9-27b6-47cd-820d-d7ecc76cbc06

https://doi.org/10.1145/3240323.3240361

Association for Computer Machinery, New York, NY

978-1-4503-5901-6

RecSys '18: Proceedings of the 12th ACM Conference on Recommender Systems

12th ACM Conference on Recommender Systems, RecSys 2018, 2018-10-02 → 2018-10-07, Vancouver, Canada

Accepted author manuscript

Institutional Repository

conference paper

© 2018 Zhu Sun, J. Yang, J. Zhang, A. Bozzon, Long Kai Huang, Chi Xu