Vers la détection des dislocations à gauche dans les transcriptions automatiques du Français parlé / Towards automatic recognition of left dislocation in transcriptions of Spoken French

Corinna Anderson 1 Christophe Cerisara 2 Claire Gardent 1
1 TALARIS - Natural Language Processing: representation, inference and semantics
Inria Nancy - Grand Est, LORIA - Laboratoire Lorrain de Recherche en Informatique et ses Applications
2 PAROLE - Analysis, perception and recognition of speech
INRIA Lorraine, LORIA - Laboratoire Lorrain de Recherche en Informatique et ses Applications
Abstract : Left dislocations are an important distinguishing feature of spoken French. In this paper, we present a hybrid approach for detecting the coreferential link that holds between left-dislocated elements and the coreferential pronoun occurring further on in the sentence. The approach combines a symbolic graph rewrite step with a maximum entropy classifier and achieves around 40% F-score. We conjecture that developing such approaches could contribute to the general anaphora resolution task and help improve parsers trained on corpora enriched with left dislocation anaphoric links.
Liste complète des métadonnées

Cited literature [4 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-00600510
Contributor : Christophe Cerisara <>
Submitted on : Wednesday, June 15, 2011 - 9:37:07 AM
Last modification on : Friday, February 9, 2018 - 1:20:04 PM
Document(s) archivé(s) le : Friday, November 9, 2012 - 3:10:37 PM

File

disloc.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-00600510, version 1

Collections

Citation

Corinna Anderson, Christophe Cerisara, Claire Gardent. Vers la détection des dislocations à gauche dans les transcriptions automatiques du Français parlé / Towards automatic recognition of left dislocation in transcriptions of Spoken French. Traitement Automatique des Langues Naturelles - TALN'2011, Jun 2011, Montpellier, France. pp.6. ⟨hal-00600510⟩

Share

Metrics

Record views

417

Files downloads

165