Autonomous Chain-of-Thought Distillation for Graph-Based Fraud Detection

TL;DR

FraudCoT introduces an autonomous, graph-aware chain-of-thought reasoning framework combined with scalable co-training, significantly improving fraud detection accuracy and training efficiency on text-attributed graphs.

Contribution

The paper presents FraudCoT, a novel framework that enables autonomous reasoning and scalable co-training for graph-based fraud detection, overcoming limitations of existing LLM-GNN methods.

Findings

Up to 8.8% AUPRC improvement over state-of-the-art

Achieves up to 1,066x training speedup

Enhances semantic-structural understanding through CoT distillation

Abstract

Graph-based fraud detection on text-attributed graphs (TAGs) requires jointly modeling rich textual semantics and relational dependencies. However, existing LLM-enhanced GNN approaches are constrained by predefined prompting and decoupled training pipelines, limiting reasoning autonomy and weakening semantic-structural alignment. We propose FraudCoT, a unified framework that advances TAG-based fraud detection through autonomous, graph-aware chain-of-thought (CoT) reasoning and scalable LLM-GNN co-training. To address the limitations of predefined prompts, we introduce a fraud-aware selective CoT distillation mechanism that generates diverse reasoning paths and enhances semantic-structural understanding. These distilled CoTs are integrated into node texts, providing GNNs with enriched, multi-hop semantic and structural cues for fraud detection. Furthermore, we develop an efficient asymmetric co-training strategy that enables end-to-end optimization while significantly reducing the computational cost of naive joint training. Extensive experiments on public and industrial benchmarks demonstrate that FraudCoT achieves up to 8.8% AUPRC improvement over state-of-the-art methods and delivers up to 1,066x speedup in training throughput, substantially advancing both detection performance and efficiency.