Studying Socially Unacceptable Discourse Classification (SUD) through different eyes: "Are we on the same page ? "

Bruno Machado Carneiro (1) , Michele Linardi (2, 3) , Julien Longhi (4, 5, 6)

1 ENSEA - Ecole Nationale Supérieure de l'Electronique et de ses Applications
2 CY - CY Cergy Paris Université
3 ETIS - UMR 8051 - Equipes Traitement de l'Information et Systèmes
4 IUF - Institut universitaire de France
5 IDHN - Institut des Humanités numériques
6 AGORA - EA 7392 - Laboratoire AGORA

We study Socially Unacceptable Discourse (SUD) characterization and detection in online text. We first build and present a novel corpus that contains a large variety of manually annotated texts from different online sources used so far in state-of-the-art Machine learning (ML) SUD detection solutions. This global context allows us to test the generalization ability of SUD classifiers that acquire knowledge around the same SUD categories, but from different contexts. From this perspective, we can analyze how (possibly) different annotation modalities influence SUD learning by discussing open challenges and open research directions. We also provide several data insights which can support domain experts in the annotation task.

Keywords: SUD Classification; Machine Learning; Deep Learning; Transfer Learning; Annotation Guidelines; SUD Classification Machine Learning Deep Learning Transfer Learning Annotation Guidelines

Domains: Computer Science [cs]; Linguistics

Bruno Machado Carneiro, Michele Linardi, Julien Longhi. Studying Socially Unacceptable Discourse Classification (SUD) through different eyes: "Are we on the same page ? ". International Conference on CMC and Social Media Corpora for the Humanities, Sep 2023, Mannheim, Germany, Germany. ⟨hal-04316521⟩