Patterns of fraud detection using coupled Hidden Markov Model

The Financial Services Authority does fraud detection through several activities that are recorded in the event logs for detecting fraud. Patterns of Fraud Detection are used to analyze the performances of fraud detection and predict the next fraud detection. Patterns of Fraud Detection can be observed using a map model of fraud detection. On the other hand, modeling fraud detection is difficult because the fraud detection cannot be directly observed through an event log. The event log only records activities triggering by fraud detection. This paper proposes an intention mining method for modeling fraud detection using Coupled Hidden Markov Model. The proposed method determines strategies utilizing the activities and forms a map model of fraud detection using probabilities of Coupled Hidden Markov Model. The experiment outcomes show that the proposed method gets an appropriate map model of fraud detection. This paper also demonstrates that an obtained model using proposed method gets the better validity than an obtained model using Map Miner Method.