Research And Application Of Knowledge Graph In The Field Of Financial System Anti-Fraud

With the Internet technology bringing convenience to the financial system,criminals are also using networking technology to implement more forms of financial fraud,of which credit card fraud is a high concern of banks.Traditional financial anti fraud systems are basically based on a combination of expert experience and machine learning algorithms.They manually summarize and summarize various features of financial fraud,and then conduct detection and screening.When faced with financial fraud behaviors that are more diverse,technologically advanced,and more team oriented,they are particularly challenging.Therefore,it is increasingly necessary to build a more efficient and accurate credit card anti fraud system.This article introduces the concept of knowledge graph and proposes a more efficient and accurate anti fraud analysis method.The main research content of this article is as follows:First,because the black sample data set of credit card fraud is unbalanced,it is impossible to accurately and effectively analyze the relationship characteristics of fraud individuals and high-risk group characteristics.From the perspective of knowledge graph,we can use the inherent correlation between things to find the target entities associated with the ID number,device,IP,and phone of the black sample.If the more target entities are associated with the black sample,the closer the group relationship is,and the more obvious the group characteristics are,namely the association network feature of "good people dispersed,bad people gathered".Therefore,we extend the black label data obtained by the business to obtain a correlation network dataset of actual scenarios with black sample data participation.Then,by community partitioning the obtained dataset,we can obtain various types of target risk groups that we want to analyze,and finally obtain the group network characteristic indicators of the risk groups in the correlation network,And the relationship characteristic indicators of individuals in the group.We build a credit card fraud system based on a knowledge graph,with risk group indicators and indicator acquisition methods as the core.Firstly,a knowledge graph is constructed for the entire credit card application data and third-party data,followed by group partitioning,group indicator calculation,group scoring,group risk ranking,and finally outputting high-risk group information.At the same time,in the association network of knowledge graphs,individual relationship feature indicators are also a very important type of feature data,which can be used as machine learning data for model training and improving the accuracy of machine learning models.