Génération automatique de billets journalistiques : singularité et normalité d'une sélection

Abstract : Natural language generation is used to describe, in natural language, the data answering a user query : weather forecast, elections or sport results... In most cases, the generated text do not show the singularity nor the normality of the selected data compared to the whole. Most of the time, the knowledge needed to express these particular kind of information is available. It is thus possible to detect particular exceptional or banal features of a selection and to generate automatically a text that exposes these interresting facts. This paper presents a tool aiming at specifying generators of texts containing comparisons of the selected data to the whole set. Singularity and banality exprimable using the tool are showed thanks to an example (election results). Reification of the needed knowledge (models, language ressources,... ) aims for genericity of the approche and easy reuse for other domains (weather forecast, sports,...). The prototype Summy was built to validate the approach and to demonstrate how particularity or normality of a subset of data compared to the whole can be automatically expressed.
Document type :
Conference papers
Liste complète des métadonnées

https://hal.archives-ouvertes.fr/hal-01491520
Contributor : Cyril Labbé <>
Submitted on : Friday, March 17, 2017 - 8:47:20 AM
Last modification on : Monday, February 11, 2019 - 4:36:02 PM
Document(s) archivé(s) le : Sunday, June 18, 2017 - 12:20:17 PM

File

summy_ws.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01491520, version 1

Citation

Jérémy Vizzini, Cyril Labbé, François Portet. Génération automatique de billets journalistiques : singularité et normalité d'une sélection. Extraction et Gestion des Connaissances (EGC) 2017 Atelier Journalisme Computationnel,, Jan 2017, Grenoble, France. Revue des Nouvelles Technologies de l'Information EGC 2017, 2017. 〈hal-01491520〉

Share

Metrics

Record views

248

Files downloads

154