W-HGAD

Official implementation of W-HGAD: a Wasserstein-based heterogeneous graph neural network for uncertainty-aware anomaly detection on graphs.

About

This repository contains the implementation of the W-HGAD model as described in our paper. W-HGAD is designed for uncertainty-aware anomaly detection on heterogeneous graphs, with a focus on the PolitiFact dataset.

Contents

• data/: Directory containing the PolitiFact dataset
• W-HGAD_PolitiFact.py: Python script implementing W-HGAD for the PolitiFact dataset

Key Dependencies

• PyTorch
• torch_geometric
• numpy
• scikit-learn

Usage

To run the W-HGAD model on the PolitiFact dataset:

python W-HGAD_PolitiFact.py

Ensure all dependencies are installed before running the script.

Contact

For questions or issues, please open an issue in this repository.

Thank you for your interest in W-HGAD!