Graph fraud detection: the papers that matter

What it does

This repository is a curated bibliography of academic papers on fraud, anomaly, and outlier detection using graph neural networks, transformers, and LLMs. The maintainers sort papers by year and technique, linking to PDFs and code when available. They also built an interactive web dashboard for filtering the collection and a local RAG-based chatbot pre-loaded with 250 accessible papers for conversational literature review.

The interesting bit

Most “awesome” lists are static markdown graveyards; this one tries to stay alive with an interactive dashboard and a chatbot that lets you query a subset of the papers directly. It is essentially a living literature review with a search interface and a conversational layer tacked on.

Key highlights

• Covers LLM, transformer, and graph-based approaches from before 2020 through 2026.
• Links directly to paper PDFs and corresponding code repositories.
• Includes an interactive dashboard for filtering and searching the bibliography.
• Offers a local RAG-based LLM chatbot for querying 250 publicly accessible papers.
• Organizes entries by venue and year, spanning KDD, NeurIPS, IEEE S&P, and others.

Verdict

Worth bookmarking if you are a researcher or practitioner trying to keep pace with graph-based fraud detection. Skip it if you are looking for a single open-source framework to deploy—this is a map, not a toolkit.