Research And Implementation Of Education Knowledge Graph Management System Based On Entity Enhancement

In recent years,with the continuous development of Internet technology,the demand for online learning is more and more widespread.Although the Internet can access a variety of learning materials,but at the same time learning resources are confusing and complex and the quality of the shortcomings are also obvious.With the continuous development of artificial intelligence,big data and other information technology industries,it has gradually become a common consensus to use knowledge graph related technology to solve the problem of low learning efficiency.Knowledge graph management system can effectively manage and visually display subject knowledge graph.Therefore,this paper studies mathematics subject knowledge graph management system to provide relevant help for students’ learning and improve learning efficiency and performance.The core of knowledge graph management system is to construct subject knowledge graph.Due to the polysemy of the word,such problems as unsatisfactory use effect and poor experience appear in the system after the construction of knowledge graph,and the entity disambiguation of knowledge graph becomes particularly important.Due to multiple problems such as insufficient semantic information in short text,incomplete expression of relation between words and lack of effective use of information in knowledge graph,the disambiguation of entity link in short text has great limitations.In view of this,this paper proposes an entity enhancement model,which uses the structure of candidate entity item nodes to concatenate nodes and relations directly adjacent to candidate entity item nodes into corresponding statements,and the text composed of all nodes and relations serves as the information supplement of candidate entity item nodes.In this method,the candidate entity items and the corresponding extended text information are used as the input of BERT pre-training model.By improving BERT feature output and using the self-attention mechanism and Textcnn convolutional neural network model,the specific information in the text is extracted as the features of entity disambiguity.The experimental results show that the proposed method can enhance the syntactic and semantic features of the candidate entity items,and play a certain role in improving the effect of entity disambiguation.The front end uses VUE to realize the visual interface.The back end uses Spring boot to process the business process.Neo4 j and Mysql are used to store entity relationship data and user data respectively.Through the user authentication module to authenticate the user’s identity,jump to different operation interface.Subject knowledge data forms knowledge map by constructing modules.In the knowledge graph management module,the entity and relationship can be changed and transformed,and the data can be updated to expand the knowledge graph.The visualization module is used to display the knowledge triples so as to obtain the whole and partial overview of the discipline knowledge structure.