Construction Of Subject Knowledge Graph For AI Education

In recent years,the application of advanced science and technology such as big data and artificial intelligence to the education sector has an obvious development.The traditional knowledge dissemination model,which used teachers to teach and students to learn as the main learning method,is rapidly evolving toward the AI education model.However,with the continuous increase of massive educational data,the data cannot be effectively organized and managed,utilized and shared,searched and understood,which seriously hinders the development of AI education.Therefore,because of this problem,this thesis proposes to apply the knowledge graph to the field of AI education,and constructs a set of methods from low-level data collection to disciplinary knowledge graph construction and integration,and applies it to the application of high-level intelligent knowledge search and question and answer,then it makes a practical exploration for the further development of educational informatization.Finally,the validity of the proposed method is verified by experiments.The main research contents of this thesis include the following three points.(1)Firstly,because of the field of wisdom education,by studying the construction method of the subject knowledge graph,combining the hierarchical nature of subject knowledge and the characteristics of the relationship between learning before and after,the general frame diagram of subject knowledge graph construction is proposed.It mainly includes two aspects:extraction of the entity,extraction of the relation between entity,and construction of ontology.First of all,this thesis uses entity recognition tools to identify entities.To reduce manual annotation and achieve a more effective extraction of relationships between entities,the remote supervised relationship extraction model of CNN+Attention is proposed.Also,to construct ontology to guide the construction of subject knowledge graph,this thesis proposes a method of keyword extraction to obtain the subject concept set and uses the instance filling algorithm proposed in this thesis to connect the extracted discipline knowledge to the ontology,thus completing the construction of the discipline knowledge graph.(2)Also,to construct a more accurate subject knowledge graph,the overall framework of subject knowledge graph fusion is proposed in this thesis,which includes three parts:data preprocessing,similarity calculation,and knowledge fusion.Firstly,the knowledge subgraph is reduced to remove redundant data.Besides,the similarity calculation method of the knowledge subgraph is proposed to calculate its similarity.Finally,the proposed matrix fusion algorithm is used to realize the fusion of the subject knowledge graph by combining the structural information of the subject knowledge subgraph.(3)In the end,to apply the constructed subject knowledge graph,this thesis constructs a set of subject knowledge search and knowledge question answering systems,which further realizes the high-level application of the knowledge graph of wisdom education.The comprehensive experimental results show that,compared with the general knowledge graph construction methods,the proposed method has a better performance in terms of construction efficiency and accuracy,and fully proves that the proposed method can realize the efficient and rapid construction of knowledge graph within the discipline domain.The research results obtained in this thesis provide a reference for the management and application of big data of educational knowledge,to promote the further development of intelligent education.