Étiquetage morpho-syntaxique pour des mots nouveaux

Abstract : Part-of-speech (POS) taggers are more or less robust with respect to the labeling of unknown words not found in the training corpus. It is important to know precisely how these tools perfom when we target part-of-speech tagging for formal neologisms. Indeed, grammatical category is an important criterion for both their identification and documentation. We present an evaluation and comparison of 7 POS taggers for French, based on a corpus built from Wiktionary. The results show that the use of form-related or morphological features supports the accurate tagging of new words.
Document type :
Poster communications
Liste complète des métadonnées

https://hal.inria.fr/hal-00998866
Contributor : Ingrid Falk <>
Submitted on : Friday, July 18, 2014 - 10:45:52 PM
Last modification on : Thursday, February 7, 2019 - 2:59:55 PM
Document(s) archivé(s) le : Thursday, November 20, 2014 - 4:08:21 PM

Files

Identifiers

  • HAL Id : hal-00998866, version 1

Collections

Citation

Ingrid Falk, Delphine Bernhard, Christophe Gérard, Romain Potier-Ferry. Étiquetage morpho-syntaxique pour des mots nouveaux. Brigitte Bigi. 21ème conférence sur le Traitement Automatique des Langues Naturelles, Jul 2014, Marseille, France. 21ème Traitement Automatique des Langues Naturelles, pp.431, 2014. ⟨hal-00998866⟩

Share

Metrics

Record views

516

Files downloads

1158