DHGE

Official resources of "DHGE: Dual-View Hyper-Relational Knowledge Graph Embedding for Link Prediction and Entity Typing". Haoran Luo, Haihong E, Ling Tan, Gengxian Zhou, Tianyu Yao, Kaiyang Wan. AAAI 2023 [paper].

Overview

An example of DH-KG structure:

Overall DHGE model:

Requirements

This project should work fine with the following environments:

• Python 3.7.11 for training & evaluation with:
  • Pytorch 1.8.1+cu101
  • numpy 1.20.3
• GPU with CUDA 10.1

All the experiments are conducted on a single 11G GeForce GTX 1080Ti GPU.

How to Run

Unzip datasets

unzip -o -d dataset/ dataset/JW44K-6K.zip
unzip -o -d dataset/ dataset/HTDM.zip

Training & Evaluation

To train and evaluate the DHGE model for tasks of link prediction and entity typing on JW44K-6K dataset, please run:

python run.py

To train and evaluate the DHGE model for tasks of medicine prediction and medicine class prediction on HTDM dataset, please run:

python run_med.py

BibTex

If you find this work is helpful for your research, please cite:

@article{luo2023dhge, 
  title={DHGE: Dual-View Hyper-Relational Knowledge Graph Embedding for Link Prediction and Entity Typing}, 
  volume={37}, 
  url={https://ojs.aaai.org/index.php/AAAI/article/view/25795}, 
  DOI={10.1609/aaai.v37i5.25795}, 
  author={Luo, Haoran and E, Haihong and Tan, Ling and Zhou, Gengxian and Yao, Tianyu and Wan, Kaiyang}, 
  year={2023}, 
  month={Jun.}, 
  pages={6467-6474} 
}

For further questions, please contact: luohaoran@bupt.edu.cn.