Alida, a cognitive approach of text categorization

Abstract

This paper proposes a model of text categorization named Alida, which combines a model of categorization inspired of the classical cognitive models of categorization of Nosofsky, with a semantic space model as system of semantic knowledge representation. The model addresses large-scale text categorization applications in opinion mining in different domains and different languages. The performance in the text-mining campaign DEFT'09 shows that the model can compete with existing Natural Language Processing and Information Retrieval models.