Pytorch and Numpy are needed to run the code.
pip install -r requirements.txt
Data can be obtained from here.
To train CORE for different datasets and different embedding schemes, run:
bash [Dataset Name]_[Embedding Model]_train.sh
We run all experiments with NVIDIA V100 GPU with 32GB memory. CUDA 10.1 or above needs to be installed.
We refer to the code of RotatE. Thanks for their contributions.
Citation
If you find the source codes useful, please consider citing our paper:
@article{ge2022core,
title={CORE: A knowledge graph entity type prediction method via complex space regression and embedding},
author={Ge, Xiou and Wang, Yun-Cheng and Wang, Bin and Kuo, CC Jay},
journal={Pattern Recognition Letters},
volume={157},
pages={97--103},
year={2022},
publisher={Elsevier}
}