Asymmetric Neighboring Context Modeling for Knowledge Graph Embedding

Abstract

Knowledge graph embedding (KGE) is to learn how to represent the low dimensional vectors for entities and relations based on the observed triples. When dealing with surrounding information, recent models either ignore the interactions between triples within the knowledge graph or use too many parameters to take the surrounding information into the model. Besides, the asymmetric information in the surrounding triples deserves further investigation. In this paper, we propose an Asymmetric Context Aware Representation for Knowledge Graph Embedding method (AcarE). Specifically, we first use an asymmetric context encoder to introduce the surrounding triples information to the head and relation entity. Afterwards we use an encoding system based on convolutional neural network (CNN) to encode the context-aware head and context-aware relation. Experimental results on both WN18RR and FB15K-237 datasets demonstrate the AcarE’s promising potential.