Bridging Data Science and Compliance Intelligent AML Systems by Design: Domain-Aware Machine Learning
DOI:
https://doi.org/10.63282/3050-9416.IJAIBDCMS-V6I4P122Keywords:
Anti-Money Laundering (AML), Domain-Aware Machine Learning, Compliance Automation, Financial Crime Detection, Explainable AI (XAI), Natural Language Processing (NLP), Anomaly Detection, Regulatory Technology (Regtech), Investment Banking, Model GovernanceAbstract
The rapid increase in global financial transactions has made the old, rules based, Anti-Money laundering (AML) systems ineffective and efficient. Such systems produce too many false positives, are expensive to operate, and fail to keep pace with more advanced and progressive typologies of crime. The paper suggests an innovative architecture of the next generation AML systems by integrating the powerful machine learning (ML) with deep compliance knowledge domain. We support a domain-sensitive strategy in which the development of the ML model, which includes feature engineering and natural language processing (NLP), pattern recognition and anomaly detection, is domain-directed, meaning that it is driven by the knowledge in the financial crime field, regulatory reporting, and operational specificity of investment banking. The paper examines the technical architectures that combine unsupervised learning to detect anomalies, supervised models to detect typology, and NLP to generate alert narrative automatically and analysis of unstructured data. More importantly, it deals with the extremely important issues of model explainability and adversarial robustness and the combination of both with the current compliance processes in order to make them regulatory-acceptable and effective. This paradigm will increase the precision of detection, minimize the workload on investigators, and bring a more intelligent and lively approach to financial crime defense through the combination of the technical profundity of data science with the fine-tuning of compliance logic
References
1. Hanif, R., Ahmad, H. S., & Ali, A. (2025). Developing an Integrated AML Risk Management Framework for Commercial Banks Based on Customer Risk Profiling and Enhanced Due Diligence. Advance Journal of Econometrics and Finance, 3(3), 206-215. http://ajeaf.com/index.php/Journal/article/view/114
2. Ofoegbu, K. D. O., Osundare, O. S., Ike, C. S., Fakeyede, O. G., & Ige, A. B. (2024). Proactive cyber threat mitigation: Integrating data-driven insights with user-centric security protocols. Computer Science & IT Research Journal, 5(8), 2083-2106. https://www.researchgate.net/profile/Kingsley-Ofoegbu/publication/383606826_Proactive_cyber_threat_mitigation_Integrating_data-driven_insights_with_user-centric_security_protocols/links/66d3512c2390e50b2c21f33b/Proactive-cyber-threat-mitigation-Integrating-data-driven-insights-with-user-centric-security-protocols.pdf
3. Liu, H., Li, Y., & Wang, H. (2025). Genomas: A multi-agent framework for scientific discovery via code-driven gene expression analysis. arXiv preprint arXiv:2507.21035. https://arxiv.org/abs/2507.21035
4. Frigiola, A. (2024). Supervised Contrastive Learning for Classification of Market Stock Series (Doctoral dissertation, Politecnico di Torino). https://webthesis.biblio.polito.it/31086/
5. Kang, Y., Yang, X., Wang, G., Wang, Y., Wang, Z., & Liu, M. (2025). Can large language models effectively process and execute financial trading instructions?. Frontiers of Information Technology & Electronic Engineering, 26(10), 1832-1846. https://link.springer.com/article/10.1631/FITEE.2500285
6. Yang, H., Zhou, Y., Ji, X., Liu, Z., Tian, Z., Tang, Q., & Shi, Y. (2025). Advancing Graph Neural Networks for Complex Relational Learning: A Multi-Scale Heterogeneity-Aware Framework with Adversarial Robustness and Interpretable Analysis. Mathematics, 13(18), 2956. https://www.mdpi.com/2227-7390/13/18/2956
7. Pérez, J. V., Chávez, M. R., Prieto, M. D., & Martínez, L. R. (2025). Recent progress of anomaly detection in energy applications: a systematic literature review. Anomaly Detection-Methods, Complexities and Applications: Methods, Complexities and Applications, 3. https://books.google.com/books?hl=en&lr=&id=pcKREQAAQBAJ&oi=fnd&pg=PA3&dq=Bridging+Data+Science+and+Compliance:+Intelligent+AML+Systems+by+Design:+Domain-Aware+Machine+Learning.&ots=SUN0cJPUsZ&sig=mE6Ta-_vQ2bHMkBtBzTdto-b0jw
8. Espinoza, M. (2025). The Effect of Enhancing Medicare Claims With OSINT on the Community Membership of Fraud Rings (Doctoral dissertation, Marymount University). https://search.proquest.com/openview/c08256382439ac12b574fdfa4b13dd76/1?pq-origsite=gscholar&cbl=18750&diss=y
