A Systematic Review of Transaction Graph Modelling for Anti-Money Laundering in FinTech Apps: Methods, Architectures, and Future Research Directions

Abstract

Money laundering remains one of the most critical challenges in modern financial systems, particularly within rapidly growing FinTech ecosystems. Traditional rule-based Anti-Money Laundering (AML) systems are increasingly inadequate due to the complexity, scale, and dynamism of digital financial transactions. In response, transaction graph modelling has emerged as a powerful paradigm for detecting suspicious activities by representing financial interactions as interconnected networks. This systematic review examines 30 peer-reviewed studies published between 2018 and 2023 that focus on graph-based AML detection in FinTech applications. The review categorizes methods into graph analytics, machine learning on graphs, deep learning approaches such as Graph Neural Networks (GNNs), and hybrid AI-graph architectures. It further explores system-level implementations including real-time monitoring frameworks, distributed ledger integration, and cloud-based AML pipelines. The findings reveal a significant shift from static rule-based systems to adaptive, AI-driven graph intelligence models capable of capturing hidden laundering patterns such as layering, smurfing, and circular transactions. However, challenges persist in scalability, data imbalance, false positives, and explainability of graph-based models. Additionally, real-world deployment is limited by privacy regulations and lack of standardized datasets. This study highlights emerging trends such as temporal graph modelling, federated learning for AML, and real-time streaming graph analytics. The review concludes by proposing future research directions focusing on explainable AI, privacy-preserving graph learning, and large-scale real-time AML systems tailored for FinTech environments.