LightTEA

Codes for IJCAI2023 paper “An Effective and Efficient Time-aware Entity Alignment Framework via Two-aspect Three-view Label Propagation”

Environment

python 3.8.0
tensorflow 2.11.1
keras 2.11.0
cudatoolkit 11.3.1
faiss-gpu 1.7.2

Usage

Before running the program, you need to unzip the 'data.zip' file.

On the first run, you need to use command "python cal_simt.py" to calculate the temporal similarity matrix, then use command "python LightTEA.py" to get the results. The first run may be slow because the graph needs to be preprocessed into binary cache.

For future runs, you only need to use command "python LightTEA.py" to get the results.

Acknowledgement

We refer to the code of LightEA. Thanks for their great contributions!

Citation

If you use this model or code, please cite it as follows:
@inproceedings{Cai23_lightTEA,
author = {Li Cai and Xin Mao and Youshao Xiao and Changxu Wu and Man Lan},
title = {An Effective and Efficient Time-aware Entity Alignment Framework via Two-aspect Three-view Label Propagation},
booktitle = {Proceedings of the Thirty-Second International Joint Conference on Artificial Intelligence, {IJCAI} 2023, 19th-25th August 2023, Macao, SAR, China},
pages = {5021--5029},
publisher = {ijcai.org},
year = {2023},
url = {https://doi.org/10.24963/ijcai.2023/558},
doi = {10.24963/ijcai.2023/558},
}