Skip to Main Content
In this paper the authors investigate the task of authorship attribution on very old Arabic texts that were written by ten ancient Arabic travelers. Several features such as characters n-grams and word n-grams are used as input of a SMO-SVM (i.e. Sequential Minimal Optimization based Support Vector Machine). Experiments of authorship attribution, on this text database, show interesting results with a classification precision of 80%. This research work, which represents a rare text-mining work on the Arabic language, has revealed several interesting points.
Date of Conference: 26-28 June 2012