Learning Informativeness of Opinions

Marina Sokolova

RALI, DIRO

Le 28 juin 2007 à 11 h 30

Salle 3195, Pavillon André-Aisenstadt


Text informativeness, defined as the amount of provided information, depends on context and situation. In evaluative texts, authors provide implicit information that reinforce or diminish the explicitly stated opinion. We built language features that correspond to a few types of implicit and explicit information and use these features to represent texts of product reviews. By applying Machine Learning algorithms to the texts, we studied relations between informativeness and the strength of opinions.


Pour recevoir les annonces hebdomadaires par courriel, envoyez un message à l'adresse majordomo@iro.umontreal.ca

Liste des autres séminaires