REPRESENTACIÓN CONCEPTUAL PARA LA CLASIFICACIÓN MULTILINGUE DE TEXTOS.

Nowadays, the percentage of english information available in Word Wide Web is decreasing, because other languages such as: Chinese, Spanish, Arabic and Portuguese are gaining acceptance and dissemination. This phenomenon has caused that multilingualism become as one of the major challenges for intelligent documents processing, management, and retrieval. In order to deal with this problem efficiently, computer's system need to design new models or improve traditional models for documents representation. The availability of multilingual concepts repositories and semantic networks has opened an attractive approach to model documents written in different languages as concept vectors into a common space of representation. In this paper we present a new concept-based representation using Multilingual Central Repository. Our proposals apply a coarse-grained word sense disambiguation for selecting the appropriate concept according to topic and relevant domains discussed in documents. We experimentally evaluate our proposed method into a multilingual document classifications task. The results obtained in the experiments are encouraging, and demonstrate the usefulness of the proposed method. [ABSTRACT FROM AUTHOR]