Domain Classification-based Source-specific Term Penalization for Domain Adaptation in Hate-speech Detection

Tulika Bose, Nikolaos Aletras, Irina Illina, Dominique Fohr


Abstract
State-of-the-art approaches for hate-speech detection usually exhibit poor performance in out-of-domain settings. This occurs, typically, due to classifiers overemphasizing source-specific information that negatively impacts its domain invariance. Prior work has attempted to penalize terms related to hate-speech from manually curated lists using feature attribution methods, which quantify the importance assigned to input terms by the classifier when making a prediction. We, instead, propose a domain adaptation approach that automatically extracts and penalizes source-specific terms using a domain classifier, which learns to differentiate between domains, and feature-attribution scores for hate-speech classes, yielding consistent improvements in cross-domain evaluation.
Anthology ID:
2022.coling-1.578
Volume:
Proceedings of the 29th International Conference on Computational Linguistics
Month:
October
Year:
2022
Address:
Gyeongju, Republic of Korea
Venue:
COLING
SIG:
Publisher:
International Committee on Computational Linguistics
Note:
Pages:
6656–6666
Language:
URL:
https://aclanthology.org/2022.coling-1.578
DOI:
Bibkey:
Cite (ACL):
Tulika Bose, Nikolaos Aletras, Irina Illina, and Dominique Fohr. 2022. Domain Classification-based Source-specific Term Penalization for Domain Adaptation in Hate-speech Detection. In Proceedings of the 29th International Conference on Computational Linguistics, pages 6656–6666, Gyeongju, Republic of Korea. International Committee on Computational Linguistics.
Cite (Informal):
Domain Classification-based Source-specific Term Penalization for Domain Adaptation in Hate-speech Detection (Bose et al., COLING 2022)
Copy Citation:
PDF:
https://aclanthology.org/2022.coling-1.578.pdf