Research And Application Of Construction Of Financial Knowledge Graph And Mining Technology Based On Tax Data

As the Internet has become more widely used and related technologies have advanced rapidly,the amount of data generated by various industries has grown exponentially,especially in the financial industry.Massive data and the complexity and diversity of data sources have brought great difficulties to data processing and information analysis,which has promoted the development and research of data mining technology.Due to its robust semantic expression,storage,and reasoning abilities,the knowledge graph has a broad application space in the financial field,and provides an effective solution for data knowledge organization and intelligent applications in the Internet era.The primary focus of this thesis’s research is the construction and mining of financial knowledge graph based on tax data.Through the exploration of research objects,it can aid users in comprehending the financial knowledge system,enhance the practical value of financial data,improve the quality of financial services,help discover and predict potential risks,and improve the stability of financial markets.Therefore,the research object of this thesis has important research significance.The main research contents of this thesis are:(1)Construction of financial knowledge graph based on tax data.By analyzing the real tax data and the demands for implementing the financial knowledge graph in practice,starting from the construction process of the graph,data processing,knowledge reasoning and modeling are performed on the tax data,and the financial knowledge graph is constructed using tax data as the foundation.(2)Data mining based on financial knowledge graph.Mining the abnormal behavior of enterprises to help discover potential risks.First,deeply analyze the abnormal purchase and sale behavior of financial enterprises,apply the BERT model to the information mining of tax purchase and sales data,and propose Improved scheme based on corpus enhancement.Then analyze and extract the main characteristics of the abnormal transaction behavior of enterprises,apply the abnormal point detection algorithm to the detection of abnormal transaction behavior,and experimentally demonstrate the applicability of the isolated forest algorithm in the financial and taxation scenario.Finally,it is proposed to apply the Louvain community discovery algorithm to the complex tax transaction graph network to complete the group mining using the enterprise’s transaction flow as the foundation and display the graph.(3)Design and implementation of financial knowledge graph visualization system.Sorted and analyzed the industry background and business requirements in the financial field,completed the design of the overall system framework,combined with the core framework and key components such as the graph database JanusGraph,the front-end framework Vue,and the server framework SpringBoot,to finalize the design and implementation of the visualization system.Lastly,the system’s primary functions are devised and presented.This thesis studies and completes the whole process from raw data to the construction of industry knowledge graphs,then to data mining based on knowledge graphs,and finally to the construction of visualization systems,and effectively incorporates cutting-edge technologies,such as knowledge graphs,graph database storage and computation,highavailability technology and data mining.This thesis provides multidirectional implementation solutions and improvement ideas for the construction and application of industry knowledge graphs.