A Method for Knowledge Graph Entity Alignment Based on Feature Embedding
LUAN Ruipeng
HAO Ruidong
Abstract：In recent years,knowledge graphs have been widely applied to many fields including intelligent question and an-swer,machine translation,and entity retrieval.The neighborhood structures of co-referred entities are usually non-isomorphic in different knowledge graphs,and to solve the problem of inadequate representation of heterogeneous knowledge graphs and the un-der-utilization of entity features.This paper proposes an entity alignment method based on the joint embedding of multiple features,using entity relationship features,entity attribute features and entity name features in knowledge graphs to train the embedding rep-resentation of entities.The entity-relationship path is used as the training data for the entity-relationship features to solve the prob-lem of narrow coverage of the relational triples.The experiments show that entity relationship features have a significant effect on the entity alignment task,and the richer the entity relationships in the graph the better the alignment results.The model shows a signifi-cant improvement in entity alignment results compared with the single-feature model.
Keywords：knowledge graphknowledge fusionentity alignmentpre-training model
Publication Date：2025-02-20
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：6( 403-408 )