DAG-Explainer

This is the official implement of the paper On Data-Aware Global Explainability of Graph Neural Networks.

Installation

• Clone the repository
• Create the env and install the requirements

$ git clone https://github.com/Gori-LV/DAG
$ cd DAG
$ source ./install.sh

Usage

• Download the required datasets to /data

  The candidates were generated using gSpan.

• Download the checkpoints to /checkpoints
• Run the searching scripts with corresponding dataset.

$ source ./scripts.sh

The hyper-parameters used for different datasets are shown in this script.

Examples

We provide examples on how to use DAG-Explainer on the three dataset. Run *.ipynb files in Jupyter Notebook or Jupyter Lab.

Citation

Feel free to use our code and keep up with our progress, we kindly request you to cite our work.

@article{lv2023dag,
  title={On Data-Aware Global Explainability of Graph Neural Networks},
  author={Lv, Ge and Chen, Lei},
  journal={Proceedings of the VLDB Endowment},
  volume={16},
  number={11},
  pages={3447--3460},
  year={2023},
  publisher={VLDB Endowment}
}