Zero-Shot Knowledge Graph Completion for Recommendation System

Wang, Zhiyuan1; Chen, Cheng2; Tang, Ke1

10.1007/978-3-031-21753-1_19

2022

23rd International Conference on Intelligent Data Engineering and Automated Learning (IDEAL)

0302-9743

1611-3349

978-3-031-21752-4

LECT NOTES ARTIF INT

13756

NOV 24-26, 2022

null,Manchester,ENGLAND

GEWERBESTRASSE 11, CHAM, CH-6330, SWITZERLAND

SPRINGER INTERNATIONAL PUBLISHING AG

Knowledge graphs are structured representations of actual entities and relations. They are widely used to improve the performance of downstream tasks such as recommendation systems and semantic searching. Knowledge graph completion (KGC) is a technology for discovering the missing relations between the entities in a knowledge graph (KG). Existing methods leverage known relations on a KG to build a model to predict missing relations. Such methods implicitly require a substantial number of relations to be known in advance, which might not be available in practice. To cope with the cold-start scenario for KGC, i.e., no relation is known in advance, we propose a zero-shot approach in this paper. Our approach converts the KGC process to an optimization problem. It uses the Evolutionary Strategy (ES) algorithm to optimize a model used to complete the KG according to the performance of the recommendation system constructed based on the completed KG. Experiments on a movie dataset demonstrate that our approach can complete the KG in the cold-start scenario and improve the performance of the recommendation system built based on the completed KG.

Knowledge graph completion Zero-shot Recommendation system Evolutionary strategy

First ; Corresponding

English

CPCI-S

Computer Science

Computer Science, Artificial Intelligence ; Computer Science, Information Systems

WOS:000904430900019

Web of Science

1.Southern Univ Sci & Technol, Dept Comp Sci & Engn, Guangdong Key Lab Brain Inspired Intelligent Comp, Shenzhen 518055, Peoples R China
2.Southern Univ Sci & Technol, Res Inst Trustworthy Autonomous Syst, Shenzhen 518055, Peoples R China

Wang, Zhiyuan,Chen, Cheng,Tang, Ke. Zero-Shot Knowledge Graph Completion for Recommendation System[C]. GEWERBESTRASSE 11, CHAM, CH-6330, SWITZERLAND:SPRINGER INTERNATIONAL PUBLISHING AG,2022.