Research On Smooth Intervention Of Individual Interaction Behavior And Identification Method

In recent years,with the development of Internet financial technology,offline transaction businesses have been intelligent and convenient with the help of online transaction platforms.However,fraud methods have become more diversified and technological as well.Expert rule anomaly detection methods are lagging and highly complex;group behavior detection methods are weak in measuring the difference of interaction behaviors between individuals;the existing detection of individual interaction behaviors could not be accurately identified in the behavioral disguise scenario where user identity is stolen.Based on the above problems,we have to solve the matter of accurately describe the user's interactive behavior and improve the recognition accuracy of the interactive behavior.This article develops in-depth research on this issue,and the main contributions are as follows:Firstly,this paper proposes a smooth intervention model of individual interaction behavior to solve the problem of high misjudgment by existing detection methods.This problem is the result of easily simulated or misappropriated by fraudsters because of the relative stability of the interaction behavior.This model starts from the user's historical interaction behavior,considers the behavior bias and stability,and determines the time of each user's behavior intervention according to the time-domain drift algorithm of interactive behavior.Users can form an interactive behavior pattern that is different from the stolen identity scenario by continuous system intervention.At the same time,an implementation method of the interactive behavior reconstruction system is proposed.Under the premise of ensuring that the basic business logic of the system is not damaged,the behavior of legitimate users can be smoothly changed and distinguished from the characteristics of the stolen identity scene.Experiments show that this method can significantly improve the detection effect of the model for identity disguise behavior.Secondly,it is proposed that a multi-factor behavior anomaly detection model based on the interactive behavior cycle.This model divides the interactive behavior satisfying the threshold into different behavior cycles by calculating the user's periodic fluctuation threshold.In addition,it is proposed a method to describe the maximum deviation of user interaction behavior from the benchmark.It enhances the characteristics of the behavior baseline vector while ensuring that the numerical characteristics of the behavior baseline vector are not damaged.It is verified by comparative experiments that the proposed multi-factor abnormal behavior detection model can identify abnormal behaviors more accurately in the detection of user interaction behavior abnormalities,and the misjudgment of normal behaviors is lower.Finally,an anomaly detection system is designed and implemented based on individual interaction behavior.The system collects the user's interactive browsing behavior in the business system and the calculation result of the model,uses page element replacement in the business system,and superimposes the incentive mechanism in the user's non-critical behavior path,so that the user's behavior is not mandatory in the system.Certain changes occur under sexual intervention.Moreover,we provide the interactive behavior abnormal analysis function for the user's interactive behavior characteristics,so that each interactive judgment result can be friendly and visualized on the administrator side.Also,the analyst can verify and analyze the abnormal behavior by the output of the current model result.In summary,this article designs and implements an anomaly detection system based on individual interaction behaviors for the problem of abnormal behavior detection in user identity theft scenarios.It proposes the extraction method and detection model of interactive behavior features from the perspective of guiding interactive behavior changes and interactive behavior periodic analysis,respectively.In this paper,the effectiveness of the model is verified by the interactive behavior anomaly detection system.It has good application value in the field of behavior disguise detection of electronic transactions.