Our work consists of designing and implementing algorithms for automatically building multi-level networks of criminal organizations and extracting hundreds of money-laundering risk red flags. Once a network has been generated our algorithms automatically detect the firms that have a high likelihood of being a shell company used by criminal organizations to launder money. These algorithms leverage on a large data lake containing data on millions of individuals, firms, public contracts, etc., as well as previously unstructured data composed of thousands of individuals and firms associated with money laundering cases brought to justice in the recent past. These algorithms have been further incorporated into a Decision Support System (DSS) that automatically produces reports that greatly optimize investigations.
Previous to the creation of this newly developed set of algorithms and the respective DSS the elaboration of the same type of report work would require hundreds of man-hours in relation to only a single investigation. Thus, it was unfeasible to carry out robust data-centered investigations in relation to most money-laundering cases. Our team is now able to analyze vast amounts of data and identifying relevant findings in a matter of minutes. We believe that the nationwide diffusion of this DSS can represent a considerable improvement in anti-money laundering policy.
It should be noted that this work has only been possible because of a strategical partnership with researchers and professors from the Computer Science Department of a prestigious Brazilian University.