A Review of Entity Alignment based on Graph Convolutional Neural Network
Zeshi Wang, Mohan Li, Zhaoquan Gu · 2021
Entity alignment in knowledge graph (KG) is one of the focuses of knowledge graph fusion work. The translation models represented by TransE used to be used for entity embedding. These methods embed the entities, attributes, and relationships in the KG into the vector space, and calculate the similarity between the embedding results to align the heterogeneous entities. With the emergence and development of graph neural networks, many studies now believe that the learning entity representation of graph convolutional neural networks (GCN) can be used to better complete the entity alignment task. This paper reviews the entity alignment work based on GCN. First, the entity alignment task and GCN are formally defined. Then, the current technical focus and latest progress of GCN-based KG entity alignment work are summarized. Then, the common data sets and evaluation indicators of KG entity alignment are introduced. Finally, the future direction of KG entity alignment based on GCN is prospected.