(#84 / Sess. 2) UniKER: A Unified Framework for Combining Embedding and Horn Rules for Knowledge Graph Inference
Abstract
Combine KGE and logical rules for better KG inference have gained increasing attention in recent years. Unfortunately, a majority of existing employ sampling strategies to randomly select only a small portion of ground rules or hidden triples, thus only partially leverage the power of logical rules in reasoning. In this paper, we propose a novel framework UniKER to address this challenge by restricting logical rules to be Horn rules, which can fully exploit the knowledge in logical rules and enable the mutual enhancement of logical rule-based reasoning and KGE in an extremely efficient way. Extensive experiments have demonstrated that our approach is superior to existing state-of-the-art algorithms in terms of both efficiency and effectiveness.