The Design And Implementation Of Bank Marketing Anti-fraud Monitoring And Analyzing System

With the rapid spread of the Internet,especially mobile Internet,traditional banking is facing the opportunities and challenges of digital transformation.To facilitate end-users and accumulate digital assets,many of the bank's operations have migrated from o?ine to online,and online marketing and promotion activities have become increasingly diverse.At the same time,there is a proliferation of fraudulent practices targeting bank marketing and promotion.In the face of specialized,systematic fraudulent gangs,traditional anti-fraud methods based on expert rules can be easily cracked and bypassed,resulting in greater economic losses.Therefore,banks' marketing anti-fraud systems need to merge with big data and machine learning technologies to provide security wind control support for banking business.The marketing anti-fraud monitoring and analysing system proposed in this thesis combines the fraud detection method based on user blacklist,fraud detection method based on expert rules and fraud detection method based on machine learning model from the actual situation of H Bank.Through the three layers of detection mechanism,a complete detection link for fraudulent users has been formed,greatly improving the identification rate of fraudulent users.The system further reduces the risk of misidentifying normal users as fraudulent users by adding a manual review mechanism.In addition,the system has established model management and e?ectiveness monitoring mechanisms to facilitate management control and iteration of the fraud model.At the architecture design level,based on the idea of microservice architecture for planning and design,we developed microservices such as User Blacklist Detection Management,Expert Rule Detection Management,Model Fraud Detection,Model Fraud Management,Manual Review of Fraud,Model E?ect Monitoring,and Unified Pushing of User Detection Results,forming an interconnected anti-fraud monitoring and analysing system based on microservice architecture.This system intercepts fraudulent users by interacting with data from the bank's marketing business system through gateway services.At the specific engineering implementation level,this system uses the Spring Cloud framework to build the system,uses Rabbit MQ as the message queue,and uses My SQL and Redis as the main data storage system.At the system deployment level,this system uses Docker containers as the deployment vehicle for microservices.At the algorithm and model level of classification prediction,this system selects SMOTE algorithm and XGBoost algorithm to build machine learning models.At present,some of H Bank's online marketing business systems have been integrated into the system.It has achieved good results in helping H Bank Marketing System to detect fraudulent users,greatly reducing the economic loss caused by user fraud and improving the actual e?ectiveness of online marketing activities.