# HGNN-AC

**Repository Path**: xiao-song-sinx/HGNN-AC

## Basic Information

- **Project Name**: HGNN-AC
- **Description**: 基于属性补全的异质图节点分类
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: main
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2021-12-08
- **Last Updated**: 2021-12-08

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# Heterogeneous Graph Neural Network via Attribute Completion
This repository contains the demo code of the paper:
>[Heterogeneous Graph Neural Network via Attribute Completion](https://doi.org/10.1145/3442381.3449914)

which has been accepted by *WWW2021*.
## Dependencies
* Python3
* NumPy
* SciPy
* scikit-learn
* NetworkX
* DGL
* PyTorch
## Datasets
The preprocessed datasets are available at [Baidu Netdisk](https://pan.baidu.com/s/1teLcrdVxrE1YQVU14sRJyw)(password: hgnn) or [Google Drive](https://drive.google.com/file/d/1PqUjvSViICa8yOszqDrw-j96hXVJ0MHR/view?usp=sharing).

Please extract the zip file to folder `data`.

We use the same methods as MAGNN to process the data, so you can also download datasets at [MAGNN's repository](https://github.com/cynricfu/MAGNN).
## Example
* `python run_DBLP.py --cuda`
* `python run_IMDB.py --cuda`
* `python run_ACM.py --cuda`

Please refer to the code for more parameters.
## Acknowledgements
The demo code is implemented based on [MAGNN: Metapath Aggregated Graph Neural Network for Heterogeneous Graph Embedding](https://github.com/cynricfu/MAGNN).
## Citing
    @inproceedings{hgnn-ac,
     title={Heterogeneous Graph Neural Network via Attribute Completion},
     author={Di Jin and Cuiying Huo and Chundong Liang and Liang Yang},
     booktitle = {WWW},
     year={2021}
    }