Research On Knowledge Base Completion And Authority Management Based On Graph Neural Network

In recent years,with the development of the computer field,scientific and technological data has gradually increased,and data storage forms have also become diversified.How to efficiently store and manage these scientific and technological big data has also become an urgent problem to be solved.Relying on the projects of "Hebei Province Scientific and Technological Innovation Big Data Standardization Processing and Application Development" and "Science and Technology Ontology Index Database Design and Implementation" projects,this paper constructs a scientific and technological knowledge map database.On the basis of the scientific and technological knowledge map database,it proposes a scientific and technological knowledge map database.The entity relationship prediction and authority management method of the knowledge graph library realizes the completion and authority management of the scientific and technological knowledge graph library.The main research contents of the thesis are as follows:(1)Relation completion of science and technology knowledge graph database based on graph neural networkFirst of all,according to the characteristics of big data in science and technology,a knowledge graph library of big science and technology is constructed,and the construction steps are given.Secondly,in order to solve the problems of insufficient relationship and loss of relationship between entities and entities in the current knowledge map in the field of science and technology,a method of entity relationship completion combining weighted graph convolutional neural network(WGCN)and attention mechanism is proposed.A weighted graph convolutional network is designed to capture the hidden features of nodes in the science and technology knowledge graph,and an attention layer is added to calculate the weight of the relationship between nodes and neighbor nodes in the graph,highlighting important relationships between entities and ignoring non-important relationships.Use the bag-of-words model to generate one-hot one-hot encoding for the entity node in the completed science and technology big data knowledge graph,train the deep learning model to obtain the entity node features,and finally implement the entity triples through the convolutional network model(Conv E)Link prediction.(2)Permission management model based on graph neural network classificationOn the basis of the completion of the scientific and technological knowledge map database,combined with the classification method of the graph convolutional neural network to realize the authority management of the knowledge map database.Compared with the traditional two-dimensional relational big data warehouse table or field division method of authority,this article starts from the entity relationship of the knowledge graph database,uses the multi-layer graph convolutional neural network model to classify the scientific and technological knowledge graph,and the classification results are carried out.Multi-level authority division;according to the result of authority division,a role and attribute-based access control method is proposed,and at the same time,a trust mechanism is combined to achieve a more fine-grained and safer authority access control to the science and technology knowledge graph.By using the multi-layer graph convolutional neural network model to classify the science and technology knowledge graph data,it has superior performance compared with the traditional node classification method.At the same time,combined with the role and attribute access control model,it can realize the more detailed science and technology knowledge graph.Granular and safer authority management.