A Cosine Maximization-Minimization approach for User-Oriented Multi-Document Update Summarization - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2007

A Cosine Maximization-Minimization approach for User-Oriented Multi-Document Update Summarization

Résumé

This paper presents a User-Oriented Multi-Document Update Summarization system based on a maximization-minimization approach. Our system relies on two main concepts. The first one is the cross summaries sentence redundancy removal which tempt to limit the redundancy of information between the update summary and the previous ones. The second concept is the newness of information detection in a cluster of documents. We try to adapt the clustering technique of bag of words extraction to a topic enrichment method that extend the topic with unique information. In the DUC 2007 update evaluation , our system obtained very good results in both automatic and human evaluations.
Fichier non déposé

Dates et versions

hal-01313205 , version 1 (09-05-2016)

Identifiants

  • HAL Id : hal-01313205 , version 1

Citer

Florian Boudin, Juan-Manuel Torres-Moreno. A Cosine Maximization-Minimization approach for User-Oriented Multi-Document Update Summarization. Recent Advances in Natural Language Processing, 2007, Borovets, Bulgaria. ⟨hal-01313205⟩

Collections

UNIV-AVIGNON LIA
47 Consultations
0 Téléchargements

Partager

Gmail Facebook X LinkedIn More