Data Mining Of Enterprise Tax Knowledge Graph And Application Of Visualization Platform Base On Graph Database

Under the rapid and iterative development of Internet technology,the complexity of business information data and relationships in various industries has increased exponentially,which has led to the research on efficient mining and analysis of massive data for specific industry scenarios.It is required to be able to explore the data interconnection in a relatively short period of time and tap potential regular information.As an emerging technology combining big data and graph computing,knowledge graph can well meet the needs of mining,analysis and visual expression of data associations in specific fields.The research motivation of this thesis is the knowledge graph mining and visualization application construction for the field of tax relation.The main contents of this thesis are as follows:(1)Research on deep mining algorithm of graph data based on tax relationship.Firstly,by constructing a tax relationship graph network model,the use of tax flow data to detect circular transactions is abstracted into a directed graph search loop problem,and The loop search algorithm based on depth-first traversal is deeply studied,the performance bottleneck of naive algorithm is analyzed,and a high-performance search optimization algorithm based on bidirectional traversal and multithreading is proposed.The community discovery algorithm research is applied to the tax flow transaction network to complete the group clustering community mining and display the effect.(2)By sorting out the complex business relationships and construction requirements of tax big data,combined with service development frameworks and storage architectures such as JanusGraph,SpringBoot,Vue,D3.JS,and other key technologies of graph rendering,the graph visualization architecture design and core component development based on graph database are completed.The construction and development realizes the rapid storage,retrieval and visual presentation of knowledge graphs for different business needs.Then the partial display of system construction results is completed.(3)The visualization of deep retrieval of graph data and the construction and import of graphs are the core business of the system application.The thesis conducts research on the optimization scheme of system retrieval and storage performance for real deployment and operation scenarios.From the application layer of the system architecture and the storage engine layer of the JanusGraph graph database,the performance optimization analysis and practice of the retrieval and storage core business in high concurrency and multi-task scenarios are carried out respectively.With the expansion of applications and the increase in the number of service calculations,it is necessary to expand the stand-alone version into a distributed cluster system application to implement the deployment and testing of the distributed cluster monitoring service.The experimental data of this thesis comes from the national tax supervision department.The research of graph depth loop information and community structure mining based on massive tax relationship data is conducted.On the basis of tax data mining analysis,a high-performance knowledge graph visualization application is designed and constructed based on graph database.The optimization analysis of system retrieval and storage performance is carried out in real deployment and operation scenarios.The research results of this thesis combine and utilize emerging technologies such as knowledge graph,graph data mining,database and distributed storage computing,and provide improved ideas and specific practices for the construction of knowledge graph applications in various directions.