Discovery Of Important Entities In Educational Knowledge Graph For High School And University Cross-Learning Science And Engineering

Artificial intelligence technology enables the digital transformation of education.Teachers and students are required to have the ability to acquire and utilize multidisciplinary knowledge between different stages.Hence,correlation learning between different stages’ knowledge is a classic problem in education.As a core technology of artificial intelligence,knowledge graph has been used in the field of education with the complicated big data.Specially,there are great differences in the teaching mode between high school and university.But the knowledge has strong correlation between math and science.It is necessary to find the connecting nodes and important nodes between science and engineering.Compared with the general knowledge graph,the educational knowledge graph of math and is highly professional,complex and compact.There is more obvious semantic correlation,structure level and logical reasoning among knowledge points.Therefore,how to use the knowledge graph to find the connecting nodes and important nodes between different stages is worthy of further study.In the process of finding connecting nodes,many studies use entity alignment to find correlations between different knowledge graphs.There is a close relationship between entities in the educational knowledge graph of math and science.Due to the obvious semantic relevance,structural hierarchy and logical reasoning of entities,it is more complex to fusion the semantic information and neighbor information of entity alignment.The knowledge graphs cannot be aligned with a single embedding model.At the same time,in the process of finding important nodes,many studies lack the consideration of the characteristics of education.They only rely on a single input signal or a single calculating perspective,which ignoring the influence of neighbor information and supplementary information.In view of the shortcomings of the existing research,this paper solves two main problems in the educational knowledge graph of math and science between high school and universities:(1)Entity alignment: how to find connecting nodes through making full use of semantic information and neighbor information.(2)Entity importance assessment: how to find important nodes through educational teaching characteristics and supplementary information.Aiming at the above problems,the main work of this paper includes:(1)As for entity alignment,this paper proposes an entity alignment model of educational knowledge graph based on fusion embedding.In order to better express the semantic relevance,structure hierarchy and logic inference of knowledge entities,this paper designs a model including the embedding layer,the presentation layer and the matching layer.The embedding layer obtains entity embedding vectors for semantic information and neighbor information,which is from the perspective of coarse and fine granularity through topic calculation.The presentation layer fuses and reduces the dimension of entity information by using neural network.The matching layer obtains the final entity pair by calculating the entity similarity.Based on the knowledge graphs of different stages,this paper verifies the accuracy of the proposed method by comparing with other entity alignment methods.(2)This paper proposes an entity importance evaluation model based on topology structure.Firstly,considering the characteristics of education,this paper proposes three computational dimensions of materiality evaluation,which quantifies the entity importance.Then,considering the influence of entity neighborhood,this paper acquires neighborhood information by using graph neural network.Finally,considering the redundancy and correlation among the entities,this paper get important entities by calculating and filtering the similarity.To sum up,this paper proposes an entity alignment model of education knowledge graph based on fusion embedding,and evaluates entity importance after alignment.Finally,this paper discusses the influence of entity importance in educational knowledge graph through experiments,and proves the effectiveness of the proposed method.