1944. A Systematic Review and Experimental Evaluation of Social Network Analytics for Anti-Money Laundering
Invited abstract in session MB-31: Crime Analytics, stream Analytics.
Monday, 10:30-12:00Room: 046 (building: 208)
Authors (first author is the speaker)
| 1. | Bruno Deprez
|
| Faculty of Economics and Business, KU Leuven | |
| 2. | Toon Vanderschueren
|
| Research Centre for Information Systems Engineering (LIRIS), KU Leuven | |
| 3. | Wouter Verbeke
|
| Faculty of Economics and Business, KU Leuven | |
| 4. | Bart Baesens
|
| Decision Sciences and Information Mangement, K.U.Leuven | |
| 5. | Tim Verdonck
|
| University of Antwerp |
Abstract
Money laundering presents a pervasive global challenge, placing a burden on society through the financing of illegal activities. To combat this phenomenon, the inclusion of network information in the modelling pipeline is increasingly being explored for more effective detection, exploiting the fact that money laundering necessarily interconnected parties. This evolution has resulted in a strong growth of the literature on social network analytics (SNA) for anti-money laundering (AML) in the past few years. At the same time, the increasing attention for network approaches for AML in the literature has created important challenges for researchers and practitioners. There is no clear overview of the existing work, there exist limited insights into how different methods compare, and it is not clear which approach works best in practice.The main objective of this paper is to construct a taxonomy of SNA for AML using a systematic literature review. This review is supplemented by an extensive experimental evaluation of graph learning methods, going from manual feature engineering, over shallow representation learning to graph neural networks. These are evaluated on two money laundering datasets, a crypto-currency and a proprietary transaction data set. The results highlight the strengths and weaknesses of the different graph learning methods in the specific context of money laundering.
Keywords
- Social Networks
- Machine Learning
- Finance and Banking
Status: accepted
Back to the list of papers