Text Clustring with Fuzzy Measure of Descriptors Weight

Our work consists in implementing a new two-dimensional descriptor in Text Mining. After the morphosyntaxic analysis of the words using the techniques of automatic treatment of the natural language, there is lost additional information which we will not neglect but rather put in a new dimension. This involves a rewriting of weight descriptors in documents by a new "fuzzy" measure. The application of this approach on an Arabic corpus involved a transformation of text words in a set of pairs (root, pattern) to be descriptors of our corpus. The morphosyntactic analysis gives all possibilities and not a single solution. We apply, then the Hidden Markov model morphosyntaxic post-analysis to detect the most likely based on the context of the word analysis. We show that we are able to achieve higher precision when compared to conventional Vector Space Model representation and Latent Semantic analysis in the context of Arabic Text Clustering. Keywords— Text Mining, Text Processing , Hidden Markov models , Natural language Processing, Fuzzy logic