Survey on Classification Approach for Text Categorization

Abstract

In the area of information retrieval, text categorization has recently become an active research topic. The goal of text categorization is to allot entrics from a set of prespecified categories to a document. Learning in a very high dimensional data space is a key challenge in a text categorization approach. Learning from such high dimensional features may prompt a high computational burden and may even hurt the classification performance of classifiers because of irrelevant and, redundant features. To improve the 'curse of dimensionality' issue and to speed up the learning procedure of classifiers, it is important to perform feature reduction to reduce the size of features. This paper introduces a Bayesian arrangement approach and J48 classifier for automatic text categorization utilizing class-specific features. For text categorization, has the proposed strategy chosen a specific feature subset for every class. The detectable significance of this methodology is that most feature selection criteria, for example, Information Gain (IG) and Maximum Discrimination (MD), can be effectively joined into this methodology. The J48 classifier saves the time and memory. The proposed system also uses Term weighting concept for preprocessing. These methods increase the accuracy of classification and feature selection process and improve the system performance.