/MixTEA

[EMNLP2023] MixTEA: Semi-supervised Entity Alignment with Mixture Teaching

Primary LanguagePython

MixTEA

Getting Started

Datasets

We use entity alignment benchmark datasets OpenEA which can be downloaded from OpenEA. You need to put the prepared data into /data folder.

Dependencies

  • Python 3
  • PyTorch
  • networkx==2.5.1
  • Scipy
  • Numpy
  • Pandas
  • Scikit-learn
  • Faiss

You can automatically download corresponding dependencies by following scripts:

pip install -r .\requirements.txt

Running

Note: The settings of hyper-parameters are given in /args folder.

To run MixTEA, please use the following scripts (ps: --task is an argument):

python run.py --task en_fr_15k
python run.py --task en_de_15k
python run.py --task d_w_15k
python run.py --task d_y_15k

To run 5-fold cross-validation, please use the following script:

python run_fold.py --task en_fr_15k

We also provide jupyter notebook version in MixTEA.ipynb.

If you have any difficulty or question in running code and reproducing experimental results, please email to xiefeng@nudt.edu.cn.

Acknowledgement

We refer to the codes of these repos: GCN-Align, OpenEA, MuGNN, MeanTeacher. Thanks for their great contributions!