Research And Application Of Telecommunication Fraud Detection Technology Based On Mobile Terminal Replacement

With the progress of the times and the rapid development of the economy,the telecommunication industry has entered a stage of vigorous development.At the same time,telecommunication fraud has also begun to spread and become a new social problem.In recent years,the Ministry of Industry and Information Technology has attached great importance to the prevention and blocking of telecommunication fraud,and operators at all levels have continuously conducted research on telecommunication fraud detection technology.However,most of the detection focuses on the call behavior of fraudulent users.This article analyzes telecommunication user data and proposes a telecommunication fraud detection model based on the behavior of changing mobile terminal(i.e.mobile terminal replacement)to identify telecommunication fraud numbers,and apply the model to the detection of telecommunication fraud numbers by operators to reduce telecommunication fraud.The main contributions of this paper are as follows:(1)A strategy(rule)model based on changing mobile terminal behavior is proposed.This model is mainly based on the experience of business personnel,through the hypothesis and analysis of the characteristics of changing mobile terminals,to determine the detection rules of the model.Subsequently,the detection rules of the model are optimized according to the complaints,which improves the effect of the model.After testing and verifying the telecommunication dataset,this model can effectively detect telecommunication fraud users and has high interpretability.(2)A machine learning model based on changing mobile terminal behavior is proposed.This model is mainly based on XGBoost and AdaBoost machine learning classification algorithms for training and prediction research to detect telecommunication fraud users.Through the index evaluation of the two algorithms,XGBoost model has obvious advantages in accuracy,recall rate and other evaluation criteria.(3)Apply the two proposed models to the detection of telecommunication fraud numbers in the current networks of Guangdong and Guizhou provinces,and evaluate the application effectiveness of the models from indicators such as probability and complaint rate.Finally,compare the advantages,disadvantages,and applicability of the two models,and make choices and parameter adjustments in practical applications.After a year of nationwide promotion and verification,the proportion of involved and reported phone numbers with changing mobile terminal has decreased by 18.3%,proving the effectiveness and feasibility of the two telecommunication fraud detection models proposed in this article.