Knowledge graph completion algorithm based on structure and semantic in-formation fusion
As a powerful knowledge representation tool,knowledge graph has become one of the key technologies in the field of artificial intelligence and information retrieval.Traditional knowledge graph completion algorithms rely mainly on graph structure information,but neglect the semantic information of entities and relations,which leads to inadequate understanding of complex semantic associations.In order to address this issue,this paper proposes an innovative knowledge graph completion algorithm based on the fusion of structure and semantic informa-tion,which captures the semantic information in triples by forward transfer generation of pre-trained language model.Reverse propagation is used to optimize the structure loss to reconstruct the knowledge graph structure in semantic embedding.The performance of the model is im-proved by combining the comparative learning models.Compared with the traditional text-based algorithm KG-Bert,the proposed algorithm improves hits@10 index by 7.8％.
万方数据
维普
