Joint Entity Summary and Attribute Embeddings for Entity Alignment Between Knowledge Graphs

Abstract

Knowledge Graph (KG) is a popular way of storing facts about the real world entities, where nodes represent the entities and edges denote relations. KG is being used in many AI applications, so several large scale Knowledge Graphs (KGs) e.g., DBpedia, Wikidata, YAGO have become extremely popular. Unfortunately, very limited number of the entities stored in different KGs are aligned. This paper presents an embedding-based entity alignment method. Existing methods mainly focus on the relational structures and attributes to align the same entities of two different KGs. Such methods fail when the entities have less number of attributes or when the relational structure may not capture the meaningful representation of the entities. To solve this problem, we propose a Joint Summary and Attribute Embeddings (JSAE) based entity alignment method. We exploit the entity summary information available in KGs for entities’ summary embedding. To learn the semantics of the entity summary we employ Bidirectional Encoder Representations from Transformers (BERT). Our model learns the representations of entities by using relational triples, attribute triples and description as well. We perform experiments on real-world datasets and the results indicate that the proposed approach significantly outperforms the state-of-the-art models for entity alignment.