Discourse structure annotation: Creating reference corpora

Alexandre Labadié (1) , Patrice Enjalbert (2) , Yann Mathet (2) , Antoine Widlöcher (1)

1 Equipe CODAG - Laboratoire GREYC - UMR6072
2 Equipe Hultech - Laboratoire GREYC - UMR6072

In this paper we address the problem of defining standards for discourse-level text annotation. We propose the URS metamodel as a common formalism in which existing models and models-to-be could be (re)formulated. This meta-model should permit to create standard annotated corpora in a consistent way, i.e. allowing alignment of various discourse structures. We also present the annotation platform Glozz supporting the URS metamodel. The approach is illustrated by an on-going campaign devoted to both topical and rhetorical structure of texts.

Domaines: Intelligence artificielle [cs.AI]; Traitement du texte et du document

Alexandre Labadié, Patrice Enjalbert, Yann Mathet, Antoine Widlöcher. Discourse structure annotation: Creating reference corpora. Workshop on Language Resource and Language Technology Standards - state of the art, emerging needs, and future developments, May 2010, La Valetta, Malta. 4 p. ⟨hal-01016656⟩