Apprendre et mesurer la conflictualité avec le deep learning ?

Céline Poudat (1)

1 BCL, équipe Logométrie : corpus, traitements, modèles

This paper focuses on the linguistic expression of conflicts in Wikipedia. I am currently working on a pragmatic annotation task. I concentrate on the markers of disagreement and conflict (Poudat 2018, Poudat and Ho-Dac 2019) keeping in mind an objective of operative and partially automated description. In this context, I am naturally interested in the possibility of detecting disagreement and conflict in discussions. Deep learning and in particular the convolutional model (CNN) that my research team is currently implementing in Hyperbase Web (Vanni et al. 2018a&b) is very attractive for a number of reasons: on the one hand, it will allow me to assess conflict detection in discussion sequences and threads ; and on the other hand, to enrich the description and annotation of conflicts with new patterns and regularities using the TDS index.

Mots clés: Conflits; Wikipedia as Corpus; Pragmatics; Interactions; Deep learning

Domaines: Linguistique

Céline Poudat. Apprendre et mesurer la conflictualité avec le deep learning ?. JADT 2020 - 15èmes Journées Internationales d'Analyse statistique des Données Textuelles, Jun 2020, Toulouse, France. ⟨hal-03234268⟩