AtNRE: Adversarial training for Neural Relation Extraction

This repository is the source code for the paper:

Adversarial Training for Relation Extraction

Yi Wu, David Bamman, Stuart Russell

University of California, Berkeley

Conference on Empirical Methods in Natural Language Processing (EMNLP) 2017, Copenhagen, Denmark.

For data, please refer to the references in our paper and download from the original sources of the datasets.

Original NYT dataset (paper, link)

Original NAACL dataset (paper, link)

For reproducibility of our results, here is the processed pickled data used in the code. PLEASE (1) do not distributed and (2) refer to the original data sources for either personal use or academic purpose.

The code is under BSD-3 license.