Legal system-oriented telecom fraud detection, identification and prevention

INTRODUCTION: With the development of technology, telecom fraud is appearing more and more frequently and causing more and more harm. OBJECTIVES: This paper focused on the detection, identification, and prevention of telecom fraud. METHODS: Firstly, the telecom fraud crime was analyzed, the existing legal system was explained, and some suggestions on the protection of telecom fraud were proposed at the legal level. Then, the characteristics of telecom fraud users were analyzed to point out the differences between fraud users and normal users in terms of call, message, and traffic behavior. Finally, the Boosting algorithm was used to detect and identify telecom fraud. RESULTS: The experiments found that the boosting algorithm had advantages in the detection and recognition of telecom fraud compared with the algorithms such as support vector machine and random forest algorithms. Among several boosting algorithms, the CatBoost algorithm performed the best, with an accuracy of 0.9465 and an F1 value of 0.9047. CONCLUSION: The results demonstrate the reliability of the CatBoost algorithm in detecting and recognizing telecom fraud, and it can be applied in practice.