Please download the data used for this project from the link, which includes the following files:
** hatespeech data
- datasets/hatespeech/test.tsv
- datasets/hatespeech/train.tsv
- datasets/hatespeech/valid.tsv
** biasbios raw data + labels + private attributes
- datasets/bios/emnlp_train_bios_twoclass.pickle
- datasets/bios/emnlp_dev_bios_twoclass.pickle
- datasets/bios/emnlp_test_bios_twoclass.pickle
** biasbios embeddings used to train debiased models
- datasets/bios/emnlp_train_cls_tc.npy
- datasets/bios/emnlp_dev_cls_tc.npy
- datasets/bios/emnlp_test_cls_tc.npy
@inproceedings{subramanian2021evaluating,
title={Evaluating Debiasing Techniques for Intersectional Biases},
author={Subramanian, Shivashankar and Han, Xudong and Baldwin, Timothy and Cohn, Trevor and Frermann, Lea},
booktitle={Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing},
pages={2492--2498},
year={2021}
}