Wikipedia as Multilingual Source of Comparable Corpora

This article describes an automatic method to build comparable corpora from Wikipedia using Categories as topic restrictions. Our strategy relies of the fact Wikipedia is a multilingual encyclopedia containing semi-structured information. Given two languages and a particular topic, our strategy builds a corpus with texts in the two selected languages, whose content is focused on the selected topic. Tools and corpora will be distributed under free linceses (General Public License and Creative Commons).