This is the PyTorch implementation of the Graph Neural Network-based Fraud Detector with Homophilic and Heterophilic Interactions (H2-FDetector), as described in our paper:
Fengzhao Shi, Yanan Cao, Yanmin Shang, Yuchen Zhou, Chuan Zhou, Jia Wu, H2-FDetector: A GNN-based Fraud Detector with Homophilic and Heterophilic Connections (WWW'22)
src/: includes all code scripts.data/: includes original datasets:YelpChi.zip: The original dataset of YelpChi, which contains hotel and restaurant reviews filtered (spam) and recom- mended (legitimate) by Yelp.Amazon.zip: The original dataset of Amazon, which contains product reviews under the Musical Instruments category.
config: includes the setting of parameters for two datasets.yelp.yaml: The general parameters of YelpChi.amazon.yaml: The general parameters of Amazon.
We take YelpChi as an example to illustrate the usage of repository.
# Unzip the dataset
upzip ./data/YelpChi.zip ./data/
# Move to src/
cd src/
# Convert the original dataset to dgl graph
# The generated dgl graph does not contains the features and graph structure, but also extracts the edge label.
python data_preprocess.py --dataset yelp
# Train and test the dataset
# If you want to change the parameters in training process, you can modify the corresponding yaml file in config.
python train.py --dataset yelp
Please cite the following paper if you use this code in your work:
@inproceedings{h2fdetector_webconf22,
title = {H2-FDetector: A GNN-based Fraud Detector with Homophilic and Heterophilic Connections},
author = {Shi, Fengzhao and Cao, Yanan and Shang, Yanmin and Zhou, Yuchen and Zhou, Chuan and Wu, Jia},
booktitle = {Proceedings of the ACM Web Conference 2022 (TheWebConf)},
pages = {1486--1494},
year = {2022}
}
If you have any questions about the paper or the code, please feel free to create an issue or contact Fengzhao Shi shifengzhao@iie.ac.cn. Sometimes I may not reply very quickly because of the engaged matters, but I will do it asap when I am free :)