Transfer Learning and Prediction Consistency for Detecting Offensive Spans of Text

Amir Pouran Ben Veyseh, Ning Xu, Quan Tran, Varun Manjunatha, Franck Dernoncourt, Thien Nguyen


Abstract
Toxic span detection is the task of recognizing offensive spans in a text snippet. Although there has been prior work on classifying text snippets as offensive or not, the task of recognizing spans responsible for the toxicity of a text is not explored yet. In this work, we introduce a novel multi-task framework for toxic span detection in which the model seeks to simultaneously predict offensive words and opinion phrases to leverage their inter-dependencies and improve the performance. Moreover, we introduce a novel regularization mechanism to encourage the consistency of the model predictions across similar inputs for toxic span detection. Our extensive experiments demonstrate the effectiveness of the proposed model compared to strong baselines.
Anthology ID:
2022.findings-acl.128
Volume:
Findings of the Association for Computational Linguistics: ACL 2022
Month:
May
Year:
2022
Address:
Dublin, Ireland
Editors:
Smaranda Muresan, Preslav Nakov, Aline Villavicencio
Venue:
Findings
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
1630–1637
Language:
URL:
https://aclanthology.org/2022.findings-acl.128
DOI:
10.18653/v1/2022.findings-acl.128
Bibkey:
Cite (ACL):
Amir Pouran Ben Veyseh, Ning Xu, Quan Tran, Varun Manjunatha, Franck Dernoncourt, and Thien Nguyen. 2022. Transfer Learning and Prediction Consistency for Detecting Offensive Spans of Text. In Findings of the Association for Computational Linguistics: ACL 2022, pages 1630–1637, Dublin, Ireland. Association for Computational Linguistics.
Cite (Informal):
Transfer Learning and Prediction Consistency for Detecting Offensive Spans of Text (Pouran Ben Veyseh et al., Findings 2022)
Copy Citation:
PDF:
https://aclanthology.org/2022.findings-acl.128.pdf
Software:
 2022.findings-acl.128.software.zip