A Re-evaluation of Knowledge Graph Completion Methods

Source code for ACL 2020 paper on Knowledge Graph Evaluation

Overview

Effect of different evaluation protocols on recent KG embedding methods on FB15k-237 dataset. For TOP and BOTTOM, we report changes in performance with respect to RANDOM protocol. Please refer to paper for more details.

Dependencies

• Compatible with TensorFlow 1.x, PyTorch 1.x, and Python 3.x.
• Dependencies can be installed using requirements.txt.

Usage:

• Codes for different models are included in their respective directories.
• Run proproc.sh for unziping the data.

Citation:

Please cite the following paper if you use this code in your work.

@inproceedings{sun-etal-2020-evaluation,
    title = "A Re-evaluation of Knowledge Graph Completion Methods",
    author = "Sun, Zhiqing  and
      Vashishth, Shikhar  and
      Sanyal, Soumya  and
      Talukdar, Partha  and
      Yang, Yiming",
    booktitle = "Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics",
    month = jul,
    year = "2020",
    address = "Online",
    publisher = "Association for Computational Linguistics",
    url = "https://www.aclweb.org/anthology/2020.acl-main.489",
    doi = "10.18653/v1/2020.acl-main.489",
    pages = "5516--5522"
}

For any clarification, comments, or suggestions please create an issue or contact Zhiqing or Shikhar.