Building Arabic corpora from Wikisource

Imene Bensalem, Salim Chikhi, Paolo Rosso

Abstract: This paper describes a new tool that helps extracting clean text from the Arabic Wikisource dump in order to build corpora. The tool purpose is illustrated by the generation of a subcorpus from Wikisource, which is a step towards the building of an evaluation corpus for Arabic intrinsic plagiarism detection.