A proposal for online analysis and identification of fraudulent financial transactions

摘要

Financial institutions handle with hundreds of thousands of wire transactions per day and need to ensure security and quality for their customers. Searching on predefined patterns is insufficient to identify frauds due to continuous evolution of fraudulent methods used by criminals. Systems used for this purpose are based on the application of some methods of Artificial Intelligence, neglect human process analysis and make little use of Visual Analytics (VA) techniques. Frauds detection domain involves time-oriented and multivariate aspects to identify anomalous transactions making fraud detection a difficult task. We propose the creation of a model for each customer based on his/her behavior, using techniques of identification of outliers and conducting analysis through VA to reduce the false positive rate in the identification of fraudulent financial transactions process. We apply this approach to a real Brazilian financial institution with a daily volume of more than 30 million of financial transactions. Our framework includes a hybrid approach: (1) use of unsupervised outlier detection algorithms; and (2) use of VA to support the real time human analysis with the aim of reducing the incidence of false positives. Potential fraudulent information are presented using VA techniques allowing specialists to evaluate suspicious transactions with no increase of the normal processing times. The initial results obtained sign that there are experimental evidence that our approach can overcome the performance of the fraud detection method today used at the Brazilian institution.