| アイテムタイプ |
会議発表論文 / Conference Paper(1) |
| 公開日 |
2026-02-19 |
| タイトル |
|
|
タイトル |
Enhancing Hate Speech Classifiers through a Gradient-assisted Counterfactual Text Generation Strategy |
| 言語 |
|
|
言語 |
eng |
| 資源タイプ |
|
|
資源タイプ |
conference paper |
| アクセス権 |
|
|
アクセス権 |
open access |
| 著者 |
Van Supranes, Michael
Peng, Shaowen
若宮, 翔子
荒牧, 英治
|
| 抄録 |
|
|
内容記述タイプ |
Abstract |
|
内容記述 |
Counterfactual data augmentation (CDA) is a promising strategy for improving hate speech classification, but automating counterfactual text generation remains a challenge. Strong attribute control can distort meaning, while prioritizing semantic preservation may weaken attribute alignment. We propose Gradientassisted Energy-based Sampling (GENES) for counterfactual text generation, which restricts accepted samples to text meeting a minimum BERTScore threshold and applies gradient-assisted proposal generation to improve attribute alignment. Compared to other methods that solely rely on either prompting, gradient-based steering, or energy-based sampling, GENES is more likely to jointly satisfy attribute alignment and semantic preservation under the same base model. When applied to data augmentation, GENES achieved the best macro F1-score in two of three test sets, and it improved robustness in detecting targeted abusive language. In some cases, GENES exceeded the performance of prompt-based methods using a GPT-4o-mini, despite relying on a smaller model (Flan-T5-Large). Based on our cross-dataset evaluation, the average performance of models aided by GENES is the best among those methods that rely on a smaller model (Flan-T5-L). These results position GENES as a possible lightweight and open-source alternative. |
| 書誌情報 |
en : Findings of the Association for Computational Linguistics: EMNLP 2025
p. 3529-3544,
ページ数 16,
発行日 2025
|
| 会議情報 |
|
|
|
会議名 |
The 2025 Conference on Empirical Methods in Natural Language Processing |
|
|
開始年 |
2025 |
|
|
開始月 |
11 |
|
|
開始日 |
04 |
|
|
終了年 |
2025 |
|
|
終了月 |
11 |
|
|
終了日 |
09 |
|
|
開催期間 |
2025-11-04 - 2025-11-09 |
|
|
開催地 |
Suzhou, China |
|
開催国 |
CHN |
| 出版者 |
|
|
出版者 |
Association for Computational Linguistics |
| 出版者版DOI |
|
|
関連タイプ |
isReplacedBy |
|
|
識別子タイプ |
DOI |
|
|
関連識別子 |
https://doi.org/10.18653/v1/2025.findings-emnlp.189 |
| 出版者版URI |
|
|
関連タイプ |
isReplacedBy |
|
|
識別子タイプ |
URI |
|
|
関連識別子 |
https://aclanthology.org/2025.findings-emnlp.189/ |
| 権利 |
|
|
権利情報Resource |
https://creativecommons.org/licenses/by/4.0/ |
|
権利情報 |
©2025 Association for Computational Linguistics. Materials published in or after 2016 are licensed on a Creative Commons Attribution 4.0 International License. |
| 著者版フラグ |
|
|
出版タイプ |
NA |
