Design And Implementation Of White List Filtering Subsystem In Telecommunication Fraud Interception System

With the development of advanced science and technology,the illegal collection and distribution of personal information has been more and more common.In recent years,the interception of telecom fraud users has become an urgent problem for telecom operators.The widely used methods of telecom fraud interception are mainly based on traditional machine learning theory.However,due to the variability of users' short-term call behavior and quantity difference between normal and fraud users,this method is difficult to achieve ideal result in practical application.According to the above requirements,this paper designs a filtering system to find telecom white list for fraud interception,which uses distributed batch processing technology to extract the features of long-term call behavior of users.Besides,it designs an algorithm to evaluate the long-term reliability of users and the degree of trust among users.Then,it maps the long-term call relationship among users to a topological graph,and realizes the filtering of telecom white list through Graph Attention Network so as to narrow the quantity difference between normal and fraud users.Firstly,this paper summarizes the basic theoretical knowledge and introduces related research methods.Secondly,it makes a detailed requirement analysis of the telecommunication fraud intercept white list filtering system.Thirdly,it designs the functional module architecture of this system according to the specific requirements,in which the cognitive algorithm of user reliability and trust evaluation algorithm are proposed.Then,the cognition results are used as initial value in graph model to complete the identification of white list users.After that,the system is designed and implemented in detail.Finally,the paper designs test cases in order to test this system and prove the final implementation meets the initial functional requirements.