Improving Text Classification Performance Using PCA and Recall-Precision Criteria

Persian text is usually associated with a wide range of important or useless features. This is the main reason why feature extraction process is one of the difficult tasks in the field of Persian text analysis and understanding. While few research works have focused on this problem, the aim of this paper is to introduce a novel approach for extracting the most relevant features and classification of Persian text. Experimental results show that utilizing the principle component analysis along with recall and precision criteria and employing term frequency and category relevancy factor can result in considerable improvement in running time of the classification process while accuracy and precision criteria are improved a little or are not decreased as much as affecting classification performance.