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.

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

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

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

@ARTICLE{kgeval, author = {{Sun}, Zhiqing and{Vashishth}, Shikhar and{Sanyal}, Soumya and{Talukdar}, Partha and{Yang}, Yiming}, title = "{A Re-evaluation of Knowledge Graph Completion Methods}", journal = {arXiv e-prints}, keywords = {Computer Science - Computation and Language}, year = "2019", month = "Nov", eid = {arXiv:1911.03903}, pages = {arXiv:1911.03903}, archivePrefix = {arXiv}, eprint = {1911.03903}, primaryClass = {cs.CL}, adsurl = {https://ui.adsabs.harvard.edu/abs/2019arXiv191103903S}, adsnote = {Provided by the SAO/NASA Astrophysics Data System} }

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