In recent years vast amount of textual information is collected and stored in various databases around the world, including the Internet as the largest database of all. Text has turned into an important and critical resource for discovering information. From business to medical and security worlds, for catching up with competitors, to treat diseases and to prevent threats against national security, necessity to extract information from the text or to establish textual associations is undeniable. This burgeoning growth of published text means that even the most avid reader cannot hope to keep up with all the reading in a field, and nuggets of insight or new knowledge are at risk of languishing undiscovered in the literature. Text mining offers a solution to this problem by replacing or supplementing the human reader with automatic systems undeterred by the text explosion. It involves analyzing a large collection of documents to discover previously unknown information. The information might be relationships or patterns that are buried in the document collection and which would otherwise be extremely difficult, if not impossible, to discover. Text clustering is one of the most important areas in text mining which includes text preprocessing, dimension reduction by selecting some terms (features) and finally clustering using selected terms. Feature selection is appeared to be the most important step in the process. Conventional unsupervised feature selection methods define a measure of discriminating power of terms to select proper terms from corpus. Evaluation of terms in groups has not been investigated in reported works. In thesis a new and robust unsupervised feature selection approach is proposed that evaluates terms in groups. Considering terms in group is to find terms that can group with other terms to be used for clustering based only on low power of discriminating. In addition a new Modified Term Variance measuring method is proposed for evaluating group of terms. Furthermore a genetic based algorithm is designed and implemented for computing new measure and finding final feature vector use by clustering task. In order to evaluate and justify our approach the proposed method and also conventional term variance method are implemented and tested using corpus collection Reuters-21578. Results of comparing these two methods are very promising and show that our method produces better average accuracy and F1-measure than conventional term variance method. .
