This is a heavily modified fork of OpenKE. This repository contains implementations of XKE-PRED and XKE-TRUE, as used in the paper:

• Arthur Colombini Gusmão, Alvaro Henrique Chaim Correia, Glauber De Bona, and Fabio Gagliardi Cozman. Interpreting Embedding Models of Knowledge Bases: A Pedagogical Approach. Proceedings of the ICML Workshop on Human Interpretability in Machine Learning, pp. 79-86, 2018. Link

This code is still under development and backward compatibility for past commits is not to be expected. A few hardcoded paths are still present in some files, but they can easily be changed to adapt the framework to your environment.

This code was built from this commit of OpenKE, which is distributed under the MIT License. All code developed thereafter is distributed under the terms of the GNU General Public License, version 3 (or, at your choosing, any later version of that license). You can find the text of that license here. This code also has PRA as a dependency, which is also licensed under the GNU General Public License.

Examples for running each stage of both XKE-PRED and XKE-TRUE can be found in the examples/ directory:

• examples/emb_grid_search/ Learn embedding models.
• examples/extract_features Extract features.
• examples/explain Learn an interpretable classifier (that uses features as input) and get its explanations.