Measure of fuzzy presence of descriptors on Arabic Text Mining

Abstract

In the present work, we propose a new model of radical descriptors in Arabic Text Mining. This model will be based on the addition of lexical information contained in the morphological pattern of the Arabic word. We developed a statistical model by Hidden Markov Chain to disambiguate the morphological analysis of corpora, and we propose a new method to measure the relationship between descriptors based on a notion of “fuzzy measure of the presence” and we adapt the traditional statistical measures to this context, and we outline the key measures of similarity and distances used in Text Mining.