Feature Preparation in Text Categorization

Text categorization is an important application of machine learning to the field of document information retrieval. Most machine learning methods treat text documents as a feature vectors. We report text categorization accuracy for different types of features and different types of feature weights. The comparison of these classifiers shows that stemmed or un-stemmed single words as features give better classifier performance compared with other types of features, and LOG(tf)IDF weight as feature weight gives better classifier performance than other types of feature weights.