On the Interaction of Identity Hate Classification and Data Bias

Donnie Parent, Nina Georgiades, Charvi Mishra, Khaled Mohammed, Sandra Kübler


Abstract
Hate speech detection is a task where machine learning models tend to be limited by the biases introduced by the dataset. We use two existing datasets of hate speech towards identity groups, the one by Wiegand et al. (2022) and a reannotated subset of the data in AbuseEval (Caselli et al. 2020). Since the data by Wiegand et al. (2022) were collected using one syntactic pattern, there exists a possible syntactic bias in this dataset. We test whether there exists such a bias by using a more syntactically general dataset for testing. Our findings show that classifiers trained on the dataset with the syntactic bias and tested on a less constrained dataset suffer from a loss in performance in the order of 20 points. Further experiments show that this drop can only be partly attributed to a shift in identity groups between datasets.
Anthology ID:
2025.ranlp-1.103
Volume:
Proceedings of the 15th International Conference on Recent Advances in Natural Language Processing - Natural Language Processing in the Generative AI Era
Month:
September
Year:
2025
Address:
Varna, Bulgaria
Editors:
Galia Angelova, Maria Kunilovskaya, Marie Escribe, Ruslan Mitkov
Venue:
RANLP
SIG:
Publisher:
INCOMA Ltd., Shoumen, Bulgaria
Note:
Pages:
900–906
Language:
URL:
https://aclanthology.org/2025.ranlp-1.103/
DOI:
Bibkey:
Cite (ACL):
Donnie Parent, Nina Georgiades, Charvi Mishra, Khaled Mohammed, and Sandra Kübler. 2025. On the Interaction of Identity Hate Classification and Data Bias. In Proceedings of the 15th International Conference on Recent Advances in Natural Language Processing - Natural Language Processing in the Generative AI Era, pages 900–906, Varna, Bulgaria. INCOMA Ltd., Shoumen, Bulgaria.
Cite (Informal):
On the Interaction of Identity Hate Classification and Data Bias (Parent et al., RANLP 2025)
Copy Citation:
PDF:
https://aclanthology.org/2025.ranlp-1.103.pdf