Subjective $\textit{Isms}$? On the Danger of Conflating Hate and Offence in Abusive Language Detection

arXiv:2403.02268v1 Announce Type: new
Abstract: Natural language processing research has begun to embrace the notion of annotator subjectivity, motivated by variations in labelling. This approach understands each annotator's view as valid, which can be highly suitable for tasks that embed subjectivity, e.g., sentiment analysis. However, this construction may be inappropriate for tasks such as hate speech detection, as it affords equal validity to all positions on e.g., sexism or racism. We argue that the conflation of hate and offence can …

abstract analysis arxiv begun cs.ai cs.cl cs.cy danger detection embed labelling language language processing natural natural language natural language processing notion processing research sentiment sentiment analysis tasks type view