Research On The Construction And Visual Interaction Of Data Structure Knowledge Graph

With the rapid development of information technology,artificial intelligence technology is widely used in the field of education,and a large number of digital education resources have emerged on the Internet.At the same time,educational resources are accompanied by the phenomenon of knowledge dispersion and the lack of association between knowledge.The problems mentioned above make people feel confused when they face large educational resources on the Internet.Fortunately,the subject knowledge graph is an excellent knowledge organization and representation technology,which can represent huge amounts of subject knowledge with diverse types and complex relationships.However,there are still some problems in the research of data structure knowledge graph,such as over-reliance on manual construction methods,varying quality of knowledge graphs,and lack of convenient methods of interaction between users and knowledge graphs.Therefore,this thesis researches the construction of the data structure knowledge graph and its visual interaction system.The contributions of this thesis are as follows.1.This thesis proposes a semi-automatic construction method for data structure knowledge graph.By combining top-down and bottom-up methods for knowledge graph construction,rule-based pattern matching and deep learning methods are used to extract and process knowledge.In addition,the results of each link are manually checked to ensure the quality of construction.This method improves building efficiency and reduces building costs.The data structure knowledge graph has the characteristics of clear ontology,clear knowledge level and good quality.2.This thesis proposes an entity recognition method with data relabeling for the construction of data structure knowledge graph.Faced with the shortage of entity recognition datasets in the data structure subject and the insufficient effect of entity recognition models trained by existing datasets,a data relabeling method is proposed to relabel entities in the corpus of existing datasets.Furthermore,the ERNIE-Bi LSTM-CRF model with excellent performance is used for entity recognition.Experimental results show that the recognition effect is improved.3.This thesis presents the design and implementation of the visual interaction system of the subject knowledge graph.The front-end and back-end development technology of Java Web is used to realize the function of browsing,querying and operating the knowledge graph,providing a more convenient editing mode for the knowledge graph.It promotes the two-way data interaction between the knowledge graph and users and further plays the role of the knowledge graph in the subject area.