safe-graph/graph-fraud-detection-papers

Combines interactive dashboards with a RAG-based LLM chatbot for exploring 250+ papers, enabling both browsing and conversational discovery. Organizes deep learning graph approaches (GNNs, Transformers) alongside traditional graph algorithms, spanning fraud detection in transactions, social networks, and tabular anomaly detection. Integrates LLM-enhanced detection methods with classical graph neural network architectures, supporting researchers across PyTorch, TensorFlow, and domain-specific fraud detection frameworks.

1,794 stars. Actively maintained with 2 commits in the last 30 days.