Adapt a Text-Oriented Chunker for Oral Data: How Much Manual Effort is Necessary?
Résumé
In this paper, we try three distinct approaches to chunk transcribed oral data with labeling tools learnt from a corpus of written texts. The purpose
is to reach the best possible results with the least possible manual correction or re-learning effort.
Domaines
LinguistiqueOrigine | Fichiers produits par l'(les) auteur(s) |
---|
Loading...