Robust Hate Speech Detection via Mitigating Spurious Correlations

Kshitiz Tiwari, Shuhan Yuan, Lu Zhang


Abstract
We develop a novel robust hate speech detection model that can defend against both word- and character-level adversarial attacks. We identify the essential factor that vanilla detection models are vulnerable to adversarial attacks is the spurious correlation between certain target words in the text and the prediction label. To mitigate such spurious correlation, we describe the process of hate speech detection by a causal graph. Then, we employ the causal strength to quantify the spurious correlation and formulate a regularized entropy loss function. We show that our method generalizes the backdoor adjustment technique in causal inference. Finally, the empirical evaluation shows the efficacy of our method.
Anthology ID:
2022.aacl-short.7
Volume:
Proceedings of the 2nd Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics and the 12th International Joint Conference on Natural Language Processing (Volume 2: Short Papers)
Month:
November
Year:
2022
Address:
Online only
Venues:
AACL | IJCNLP
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
51–56
Language:
URL:
https://aclanthology.org/2022.aacl-short.7
DOI:
Bibkey:
Cite (ACL):
Kshitiz Tiwari, Shuhan Yuan, and Lu Zhang. 2022. Robust Hate Speech Detection via Mitigating Spurious Correlations. In Proceedings of the 2nd Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics and the 12th International Joint Conference on Natural Language Processing (Volume 2: Short Papers), pages 51–56, Online only. Association for Computational Linguistics.
Cite (Informal):
Robust Hate Speech Detection via Mitigating Spurious Correlations (Tiwari et al., AACL-IJCNLP 2022)
Copy Citation:
PDF:
https://aclanthology.org/2022.aacl-short.7.pdf