A Knowledge Representation Algorithm for Disciplinary Domains
With the development of education informatization,it is especially important to con-struct high-quality subject knowledge graphs.Aiming at the current knowledge graph represen-tation learning model in the field of education only utilizes the distance information between en-tities and relations,and ignores the semantic information between them resulting in inaccurate knowledge representation.A knowledge representation learning enhancement model is pro-posed.First,the model employs a relationship matrix to identify the correlations between enti-ties and uses relationship vectors to describe the relationships between entities in the subspace.Second,the head and tail vectors are projected to the relation vectors in the vector space to en-hance the interaction between relations and entities,and to strengthen the semantic relationship between entities and relations.Finally,link prediction experiments on 2 public datasets and self-constructed subject area datasets show that the model achieves large improvements in Hit@1,Hit@3,Hit@10 and MRR compared to the baseline model.
NETL
NSTL
万方数据
维普
