Smurf Detector: a Detection technique of criminal entities involved in Money Laundering

Criminals do money laundry to hide the illegitimate sources of money to show as if their money is of a legitimate source. Money laundry has many stages that money flow has to go through to finally look as if it is of a legitimate source, rule-based systems are implemented across different banks to detect structuring which is one technique of the layering stage which sophisticated criminals can evade by unsatisfying the check rules. In this work, graph database and graph data mining are to be used to overcome this limitation, the proposed technique does this by plotting the whole transactional monetary flow of entities doing money transfers between each other as one large graph database and then detecting clusters of entities interacting with each other, afterwards detection of the most influential node (intended destination) which we consider the destination to which huge amounts of money is intended to flow to (criminal‘s account) using PageRank algorithm and eventually detecting all members (Smurfs) of participated in the paths leading to that destination, a technique that would be hard to implement using traditional RDBMS in contrary to Graph DB, our results have proven correct detection of clusters as well as the final destination of the monetary flow (criminal‘s account).