Feature Selection for Text Classification Using Machine Learning Approaches

In the present scenario, millions of internet users are contributing a huge amount of data in the form of unstructured text documents. In text classification, the high dimensional feature space, noise and irrelevant information of unstructured text documents are reducing the accuracy of text classifier. The feature selection scheme is adopted to address the high dimensional feature space problem of text classification. In this proposed research, a feature selection method based on the term frequency distribution measure is deployed. We have used the Naive Bayes and SVM classifiers with two benchmark datasets (WebKB and BBC). The experimental outcome confirms that the proposed feature selection method has a better classification accuracy when compared with other feature selection techniques.