AI Paper Atlas

Research map / Graph learning / Graph self-supervised learning

DGP: A Dual-Granularity Prompting Framework for Fraud Detection with Graph-Enhanced LLMs

Yuan Li, Jun Hu, Bryan Hooi, Bingsheng He, Cheng Chen

AAAI 2026 · Graph learning › Graph self-supervised learning · approach: Graph anomaly detection · 10 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
summarize neighbors into coarse-grained text while keeping the target node fine-grained (DGP) instead of concatenating neighbor textsQwen3-8B graph-enhanced LLM · YelpReviewsAUPRC (%): 38.36 -> 48.87 (vs InstructGLM)Improved
add graph context and summarization to a standalone LLMQwen3-8B · YelpReviewsAUPRC (%): 30.45 -> 48.87Improved
add textual summarization (ablation)DGP · YelpReviews and AmazonVideoMacroF1, AUROC, AUPRCImproved
add numerical summarization (mean aggregation) (ablation)DGP · YelpReviews and AmazonVideoMacroF1, AUROC, AUPRCImproved
add Markov Diffusion Kernel metapath trimming (ablation)DGP · YelpReviews and AmazonVideoMacroF1, AUROC, AUPRCImproved
add metapath-level summarization (ablation)DGP · YelpReviews and AmazonVideoMacroF1, AUROC, AUPRCImproved
use task-agnostic instead of task-aware summarization prompts (ablation)DGP · YelpReviewsAUPRC (%): 29.02 -> 48.87Improved
use task-agnostic instead of task-aware summarization prompts (ablation)DGP · AmazonVideoAUPRC (%): 31.82 -> 34.63Improved
shrink summarization budget to 5 tokens (ablation)DGP · YelpReviews and AmazonVideoMacroF1, AUROC, AUPRCHurt
lengthen summarization budget to 80 tokens (ablation)DGP · YelpReviews and AmazonVideoMacroF1, AUROC, AUPRCHurt
use GNNs instead of MLP on heterogeneous fraud graphsbaselines · YelpReviewsAUPRC (%): 32.24 -> 42.11 (MLP vs ConsisGAD)Improved
encoding-based prompting (GraphGPT, HiGPT) instead of text-only dual granularity promptinggraph-enhanced LLMs · YelpReviewsAUPRC (%): 30.66 (GraphGPT) vs 48.87 (DGP)Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy