TeMP-TraG: Edge-based Temporal Message Passing in Transaction Graphs

2025-03-21

AI Summary

Detecting financial crimes (e.g., money laundering, fraud) in dynamic financial transaction graphs is challenging due to rich edge features, coexisting multiple edges, and temporal evolution. To address this, we propose an edge-centric temporal message-passing mechanism: it explicitly models transaction timeliness via time-decay-weighted aggregation of neighboring edge features; and introduces dynamic edge weighting and transaction graph embedding strategies compatible with graph neural network (GNN) frameworks. Evaluated across four mainstream GNN architectures, our method achieves an average 6.19% improvement in node- and edge-level classification performance. It significantly enhances detection capability for complex, evolving financial crime patterns. Moreover, it establishes a scalable and interpretable paradigm for real-time risk control on dynamic, heterogeneous transaction graphs.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community
• Reasoning under Uncertainty: Graphical Models

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

Transaction graphs, which represent financial and trade transactions between entities such as bank accounts and companies, can reveal patterns indicative of financial crimes like money laundering and fraud. However, effective detection of such cases requires node and edge classification methods capable of addressing the unique challenges of transaction graphs, including rich edge features, multigraph structures and temporal dynamics. To tackle these challenges, we propose TeMP-TraG, a novel graph neural network mechanism that incorporates temporal dynamics into message passing. TeMP-TraG prioritises more recent transactions when aggregating node messages, enabling better detection of time-sensitive patterns. We demonstrate that TeMP-TraG improves four state-of-the-art graph neural networks by 6.19% on average. Our results highlight TeMP-TraG as an advancement in leveraging transaction graphs to combat financial crime.

Problem

• Detects financial crimes using transaction graphs
• Addresses edge features and temporal dynamics
• Improves graph neural networks for fraud detection

Innovation

• Incorporates temporal dynamics into message passing
• Prioritizes recent transactions for aggregation
• Enhances graph neural networks for financial crime detection