An Efficient Method for Automatic Text Categorization

Automatic Text Categorization refers to assigning uncategorized text documents to one or more predefined categories. Texts categorization generally divided into two main sections: feature selection and learning algorithm. For Feature selection and learning algorithms techniques, various methods have been proposed. The purpose of the proposed techniques, increasing the accuracy of classification and achieve optimal performance. In this paper a hybrid method is proposed which uses Filtering feature selection technique to reduce the complexity and works on combining classifiers outputs. The proposed method is homogeneous and uses uniform classifiers with different sampling with replacement from the training set. The results show the superiority of the proposed method compare to Naïve Bayes and j48 classifier and some related works according to the criteria of accuracy, precision, recall, F1 and classification error.