Une Approche Mixte -statistique et structurelle- pour le Résumé Automatique
Résumé
Automatic multi-document summarization techniques have recently evolved into statistical methods for selecting the sentences that will be used to generate the summary. In this paper, we present a system in accordance with « State-of-the-art » — CBSEAS — that we have developped for the « Opinion Task » (automatic summaries of opinions from blogs) and the « Update Task » (automatic summaries of newswire articles and information update) of the TAC 2008 evaluation campaign, and show the interest of structural and linguistic analysis of the documents to summarize. We also present our study on news structure and its integration to CBSEAS impact.
Origine | Fichiers produits par l'(les) auteur(s) |
---|
Loading...