A Hybrid Domain-Driven Rule-Based and Explainable Artificial Intelligence Framework for Financial Fraud Detection Using Machine Learning
Dandao Xuebao/Journal of Ballistics · 11.06.2026 · 38 (2) · 421-455
DOI: 10.52783/dxjb.v38.358 ↗Kurz gefasst
Fachartikel in der Zeitschrift Dandao Xuebao/Journal of Ballistics, online erschienen am 11.06.2026. Verfasst von Hema Verma und 1 weiteren Autor:innen.
Abstract (Originalsprache)
The increasing digitalization of financial services has led to a substantial rise in the volume and complexity of financial transactions, creating new opportunities for fraudulent activities. Traditional rule-based fraud detection systems offer transparency and regulatory compliance but often struggle to identify evolving fraud patterns and generate high false-positive rates. In contrast, machine learning techniques can effectively detect complex and previously unseen fraudulent behaviors but frequently lack interpretability and explainability. To address these challenges, this study proposes a Hybrid Domain-Driven Rule-Based and Explainable Artificial Intelligence (HDR-XAI) framework for financial fraud detection. The framework integrates domain-specific fraud detection rules derived from financial regulations and expert knowledge with advanced machine learning algorithms to enhance detection accuracy and operational efficiency. The proposed architecture employs a multi-stage process consisting of rule-based filtering, feature engineering, machine learning classification, explainable AI analysis, and risk scoring. Domain-specific rules related to transaction velocity, abnormal transaction amounts, geographical inconsistencies, customer behavior deviations, and account activity patterns are utilized to identify potentially suspicious transactions before machine learning analysis. The performance of the framework is evaluated using machine learning models including Logistic Regression, Random Forest, XGBoost, LightGBM, and Deep Neural Networks. Explainable AI techniques are incorporated to improve model transparency and support regulatory compliance requirements. The findings indicate that the hybrid approach significantly improves fraud detection accuracy, precision, recall, and F1-score while reducing false-positive rates compared with standalone machine learning models. The study contributes to the growing literature on intelligent fraud detection by demonstrating how domain expertise and data-driven analytics can be effectively integrated to create robust, interpretable, and scalable fraud prevention systems for modern financial institutions.
Zitieren
APA
Verma, H., & Sharma, M. (2026). A Hybrid Domain-Driven Rule-Based and Explainable Artificial Intelligence Framework for Financial Fraud Detection Using Machine Learning. Dandao Xuebao/Journal of Ballistics, 38(2), 421-455. https://doi.org/10.52783/dxjb.v38.358
Vancouver
Verma H, Sharma M. A Hybrid Domain-Driven Rule-Based and Explainable Artificial Intelligence Framework for Financial Fraud Detection Using Machine Learning. Dandao Xuebao/Journal of Ballistics. 2026;38(2):421-455. doi:10.52783/dxjb.v38.358
BibTeX
@article{Verma2026,
author = {Verma, Hema and Sharma, Meenakshi},
title = {A Hybrid Domain-Driven Rule-Based and Explainable Artificial Intelligence Framework for Financial Fraud Detection Using Machine Learning},
journal = {Dandao Xuebao/Journal of Ballistics},
year = {2026},
volume = {38},
number = {2},
pages = {421--455},
doi = {10.52783/dxjb.v38.358}
}
EUVIMED