Home

Awesome

GFDN

[KDD 2023] Group-based Fraud Detection Network on e-Commerce Platforms

https://doi.org/10.1145/3580305.3599836

QueryOPT

The project that helps us implement (alpha, beta)-core is QueryOPT.

Quick Start

  1. Prepare the environment: 
    conda create -n gfdn python=3.8
conda activate gfdn
conda install pytorch==1.8.1 torchvision==0.9.1 torchaudio==0.8.1 cudatoolkit=10.1 -c pytorch
pip install torch-scatter==2.0.8 -f https://pytorch-geometric.com/whl/torch-1.8.1+cu101.html && pip install torch-sparse==0.6.12 -f https://pytorch-geometric.com/whl/torch-1.8.1+cu101.html && pip install torch-geometric==2.0.0
  2. download the dataset from the competition. Please download the final round dataset.
  3. Pre-process: python dataset/get_data.py
  4. Install swig
  5. Build pyabcore: sudo apt-get -y install libboost-all-dev && cd ./queryopt && ./build.sh && cd ..
  6. Start:python main.py

Note

The last four entries of customer vertex in the dataset are noise and have been ignored. Experimental results may vary slightly due to different hardware configurations.