TeAST: Temporal Knowledge Graph Embedding via Archimedean Spiral Timeline

Codes for the paper TeAST: Temporal Knowledge Graph Embedding via Archimedean Spiral Timeline accepted by the ACL 2023.

Installation

Create a conda environment with pytorch and scikit-learn :

conda create --name teast_env python=3.8
source activate teast_env
conda install --file requirements.txt -c pytorch

Datasets

python process_icews.py

python process_gdelt.py

This will create the files required to compute the filtered metrics.

Reproducing results of TeAST

In order to reproduce the results of TeAST on the four datasets in the paper, run the following commands

python  learner.py --dataset ICEWS14    --emb_reg 0.0025 --time_reg 0.01

python  learner.py --dataset ICEWS05-15 --emb_reg 0.002  --time_reg 0.1

python  learner.py --dataset GDELT      --emb_reg 0.003  --time_reg 0.003

Acknowledgement

We refer to the code of TNTComplEx and TeLM. Thanks for their great contributions!

Cite

@inproceedings{li-etal-2023-teast,
    title = "{T}e{AST}: Temporal Knowledge Graph Embedding via Archimedean Spiral Timeline",
    author = "Li, Jiang  and
      Su, Xiangdong  and
      Gao, Guanglai",
    booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
    month = jul,
    year = "2023",
    address = "Toronto, Canada",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2023.acl-long.862",
    doi = "10.18653/v1/2023.acl-long.862",
    pages = "15460--15474" 
}