On the potential of a graph attention network in money laundering detection
| DOI | https://doi.org/10.1108/JMLC-07-2021-0076 |
| Published date | 04 October 2021 |
| Date | 04 October 2021 |
| Pages | 594-608 |
| Subject Matter | Accounting & finance,Financial risk/company failure,Financial compliance/regulation,Financial crime |
| Author | Guang-Yih Sheu,Chang-Yu Li |
On the potential of a graph
attention network in money
laundering detection
Guang-Yih Sheu and Chang-Yu Li
Department of Accounting and Information Systems, College of Management,
Chang Jung Christian University, Tainan, Taiwan
Abstract
Purpose –In a classroom,a support vector machines model with a linearkernel, a neural network and the k-
nearest neighbors algorithm failed to detect simulated money laundering accounts generated from the
Panama papersdata set of the offshore leak database. Thisstudy aims to resolve this failure.
Design/methodology/approach –Build a graph attention network having three modules as a new
money laundering detectiontool. A feature extraction module encodes these input data to create a weighted
graph structure. In it, directededges and their end vertices denote financial transactions. Each directededge
has weights for storing the frequency of money transactions and other significant features. Social network
metrics are features of nodes for characterizing an account’s roles in a money laundering typology.A graph
attention module implements a self-attention mechanism for highlighting target nodes. A classification
module furtherfilters out such targets using the biased rectified linear unit function.
Findings –Resulted from the highlighting of nodes using a self-attentionmechanism, the proposed graph
attention network outperforms a Naïve Bayes classifier, the random forest method and a support vector
machines model with a radial kernel in detecting money laundering accounts. The Naïve Bayes classifier
producessecond accurate classifications.
Originality/value –This paper develops a new money laundering detection tool, which outperforms
existing methods. This new tool producesmore accurate detections of money laundering, perfects warns of
money launderingaccounts or links and provides sharp efficiency in processingfinancial transaction records
withoutbeing afraid of their amount.
Keywords Money laundering, Social network metrics, Graph attention network,
Self-attention mechanism
Paper type Research paper
1. Background
According to the Money LaunderingPrevention Act of Taiwan, blocking money laundering
crimes is one of the responsibilities of accountant professionals. Therefore, teaching the
detection of money laundering to accountant students may be emergent. In a database
course, accountant students practiced detecting a simulated money laundering typology in
which a beneficiary receives illegal money from multiple accounts. The practice used the
Panama papers data set of the offshore leak database (Panama papers, 2021). However,
students failed to obtain accurate classifications using a support vector machines model
with a linear kernel, a neural network and the k-nearest neighbors methods. The neural
network made correct classifications for only one class. The k-nearest neighbors classifier
and support vector machinesmodel with a linear kernel suspended within 10 min.
The author Chang-Yu Li has received a funding about NT$ 42000. It serves as a scholarship to him.
This scholarship has a Grant No. 109-2813-C-309-012-H No other fundings!
JMLC
25,3
594
Journalof Money Laundering
Control
Vol.25 No. 3, 2022
pp. 594-608
© Emerald Publishing Limited
1368-5201
DOI 10.1108/JMLC-07-2021-0076
The current issue and full text archive of this journal is available on Emerald Insight at:
https://www.emerald.com/insight/1368-5201.htm
To continue reading
Request your trial