Seeds Optimization for Entity Alignment in Knowledge Graph Embedding

Abstract

The embedding-based entity alignment method usually uses pre-aligned entities as seed data, and aligns the entities in different knowledge graphs through seed entity constraints. This method relies heavily on the quality and quantity of seed entities. In this paper, we use an algorithm to optimize the selection of seed entities, and select seed entity pairs through the centrality and differentiability of entities in the knowledge graph, in order to solve the problem of insufficient number of high-quality seed entities, an iterative entity alignment method is adopted. We have done experiments on dataset DBP15K, and the experimental results show that the proposed method can achieve good entity alignment even under weak supervision.