Transaction Fraud Detection Model And Method Based On Degree Of Caution

With the rapid development of mobile payment,the phenomenon of Internet electronic transaction fraud has become more and more serious,and fraud methods have changed.Therefore,the problem of how to effectively prevent the risk of transaction fraud remains severe.Since the existing anti-fraud technologies model transaction data and interaction behaviors separately,they ignore the consistency relationship between transaction behavior and interaction behavior,resulting in inaccurate characterization of users which leading to misjudgments.In response to the above problems,this article draws on the definitions and ideas of behavior and personality in psychology,and introduces the concept of transactional personality.On the basis of transaction personality,according to the user's cautious personality,the research on the consistency relationship between transaction behavior and interactive behavior is carried out,as follows:First of all,the single modeling method that separates transaction behavior and interaction behavior to describe users in existing research ignores the consistency of transaction behavior and interaction behavior,which leads to the problem of insufficient user description.This paper proposes a user identification model with transactional personality.The model first extracts cautious features from interactive behaviors and constructs a user's caution model and cautiousness benchmark.Second,extract transaction behavior characteristics from transaction data to construct a benchmark for transaction behavior.Then,the Euclidean distances from the transaction to the benchmark of caution and to the benchmark of transaction behavior are calculated,and a mapping model based on the Euclidean distance is constructed.Finally,on the basis of the mapping model,a user behavior recognition model with transaction personality is proposed.Experiments show that the overall performance of the model,compared with the detection model based only on user behavior,is improved by about 20% on average,which verifies the accuracy of the model.Secondly,by sudden changes in user behavior under the influence of stress conditions,which are different from the original behavior,the existing detection models misjudge such behaviors leading reduction of the detection accuracy.In this paper,a user identification model based on caution prediction is proposed.The model first considers the influence of external environment and personal factors on cautious personality,uses product sales and activity time to characterize activities influencing factors,and uses users' spending power to characterize users' economic pressure.Then,based on internal and external factors,the caution prediction model is proposed to predict the degree of user's caution before the transaction.Finally,by judging whether the cautious value of real-time transaction matches the predicted value,a user identification method based on the caution prediction is proposed,which realizes the identification of the user in the early stage of the transaction.Experiments have proved that compared with the detection method based on the transaction behavior certificate,the overall detection effect of this method is improved by about 20%,which verifies the ability of this method to dynamically identify users in changing scenarios.Finally,a transaction fraud detection system based on transaction personality is designed and implemented.Based on the above two models,the system realizes user fraud detection through transaction personality in general and unexpected situations.Through the simulation of transaction scenarios by real data,these functions including real-time user transaction monitoring,audit analysis and user analysis are realized.The above functions using visual means to monitor user transactions,give the judgment results and reasons for interception,which increase the readability and interpretability of the model results.By this system,the effectiveness and practicability of these models are verified.In summary,this article focuses on user research on the cautious attributes of transaction personality and related behaviors which proposes two different transaction fraud detection models and methods.A transaction fraud detection system based on transaction personality is realized.The recurrence of transaction scenarios through real data proves the effectiveness and application value of the model proposed in this paper.