Anti money laundering using a two-phase system

Purpose – This paper aims to compete and detect suspicious transactions that can lead to detecting money laundering cases. Design/methodology/approach – This paper presents a plan-based framework for anti-money laundering systems (PBAMLS). Such a framework is novel and consists of two phases, in addition to several supporting modules. The first phase, the monitoring phase, utilizes an automata approach as a formalism to detect probable money laundering. The detection process is based on a money laundering deterministic finite automaton that has been obtained from the corresponding regular expressions which specify different money laundering processes. The second phase is STRIPS-based planning phase that aims at strengthening the belief in the probable problems discovered in the first (monitoring) phase. In addition, PBAMLS contains several supporting modules for data collection and mediation, link analysis and risk scoring. To assess the applicability of PBAMLS, it has been tested using different cases studies. Findings – This framework provides a clear shift of anti-money laundering systems (AML) from depending heuristic and human expertise to making use of a rigorous formalism to accomplish concrete decisions. It minimizes the possibilities of false positive alarms and increases the certainty in decision-making. Practical implications – This framework enhances the detection of money laundering cases. It also minimizes the number of false-positive alarms that waste the investigators’ efforts and time; it decreases the efforts presented by the investigators. Originality/value – This work proposes PBAMLS as a novel plan-based framework for AML systems.