L2IR: Revealing Latent Intent in Graph Fraud Detection
2026-05-25 • Artificial Intelligence
Artificial Intelligence
AI summaryⓘ
The authors address the problem of detecting fraud in networks where fraudsters hide by connecting a lot with normal users, which makes it harder for existing methods to spot them. They propose a new method called L2IR that uses Large Language Models to figure out the hidden intentions behind user behaviors and suspicious connections. This helps the method tell real helpful links from fake ones that try to hide fraud. The approach also uses adaptive self-training to improve performance even when there are few labeled examples. Tests on real data show their method works better than existing ones and can improve other graph-based fraud detectors.
Graph Neural NetworksFraud DetectionLarge Language ModelsNeighborhood AggregationLatent IntentSelf-TrainingBehavioral TracesCamouflage in FraudGraph-based DetectorsAUPRC
Authors
Jinsheng Guo, Zhenhao Weng, Yibo Liu, Yan Qiao, Meng Li
Abstract
Graph fraud detection has long depended on Graph Neural Networks (GNNs) to propagate and aggregate information across relational data. A critical obstacle in practice, however, is that fraudsters frequently disguise themselves by forging numerous connections with benign users, causing fraud signals to be progressively diluted during neighborhood aggregation and undermining detection reliability. While recent efforts have used Large Language Models (LLMs) to provide rich semantic cues for fraud detection, the underlying intent behind suspicious connections remains insufficiently explored. Compounding this issue, the scarcity of annotated fraud samples makes it difficult to train detectors that remain robust under heavy camouflage. To address these gaps, we propose L2IR, an LLM-driven Latent Intent Revealing framework for graph fraud detection. By uncovering latent intent from both user behaviors and suspicious connections, L2IR extracts intent-aware representations from raw behavioral traces and reasons about the true purpose behind individual connections, effectively distinguishing supportive links from misleading ones. It further incorporates adaptive self-training to enhance robustness under limited supervision. Evaluations on two real-world datasets characterized by pervasive camouflage demonstrate that L2IR surpasses strong baselines and can function as a plug-in enhancement for a range of GNN-based detectors, improving AUPRC by up to 8.27%.