OverviewFRAUDAR is an algorithm for catching fraudulent blocks in graph datasets (e.g. a review graph, Twitter follow graph, etc.) in a camouflage-resistant way.
FRAUDAR has the following properties:
- Scalable: scales near-linearly with input size
- Provably Accurate: provides high accuracy in real data, with theoretical guarantees
- Camouflage-Resistant: it is designed to minimize the ability of adversaries to evade detection
FRAUDAR: Bounding Graph Fraud in the Face of Camouflage.
Bryan Hooi, Hyun Ah Song, Alex Beutel, Neil Shah, Kijung Shin, Christos Faloutsos.
Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining 2016, San Francisco, USA
CodeThe source codes used in the paper are available. [Download]
DatasetsThe Amazon, Epinions and wiki-Vote datasets used in the paper can be found at https://snap.stanford.edu/data/.
The TripAdvisor dataset can be found at http://times.cs.uiuc.edu/~wang296/Data/.
The Twitter dataset can be found at https://an.kaist.ac.kr/traces/WWW2010.html.