Skip to Main content Skip to Navigation
Conference papers

Identification automatique de chaînes de coréférences : vers une analyse des erreurs pour mieux cibler l'apprentissage

Abstract : Automatic identification of coreference chains: Towards a linguistic analysis of errors in order to improve machine learning features. We present a preliminary qualitative study dealing with the linguistic analysis of the errors made by NLP systems dedicated to the automatic detection of coreference chains. We describe several cases of noise and silence, characterized with different degrees of importance, as well as coreference-specific types of errors, for instance the construction of "catch-all" chains that group non-used referring expressions. In order to further define a generalizable methodology, we propose a first typology of errors, and some guidelines for their consideration within the machine learning process. This research implies considerations on the possible types of hybrid systems.
Complete list of metadata

Cited literature [23 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-01819602
Contributor : Frédéric Landragin Connect in order to contact the contributor
Submitted on : Wednesday, June 20, 2018 - 4:30:25 PM
Last modification on : Friday, October 15, 2021 - 1:40:08 PM
Long-term archiving on: : Tuesday, September 25, 2018 - 4:23:33 PM

File

18_ATALA.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01819602, version 1

Citation

Frédéric Landragin, Bruno Oberle. Identification automatique de chaînes de coréférences : vers une analyse des erreurs pour mieux cibler l'apprentissage. Journée commune AFIA-ATALA sur le Traitement Automatique des Langues et l’Intelligence Artificielle pendant la onzième édition de la plate-forme Intelligence Artificielle (PFIA 2018), Jul 2018, Nancy, France. ⟨hal-01819602⟩

Share

Metrics

Record views

141

Files downloads

204