Comparative Assessment of the Performance of Three WEKA Text Classifiers Applied to Arabic Text

This research is conducted in order to compare the performance of three known text classification techniques namely, Support Vector Machine (SVM) classifier, Naïve Bayes (NB) classifier, and C4.5 Classifier. Text classification aims to automatically assign the text to a predefined category based on linguistic features, and content. These three techniques are compared using a set of Arabic text documents that are collected from different websites. The document set falls into four major categories, namely, sports, economics, politics, and prophet Mohammed sayings (Al-Hadeeth Al-Shareef). The text documents pass through a set of preprocessing steps such as removing stop words, normalizing some characters, removing non Arabic text and symbols. These documents are then converted to the appropriate file format that can be used to run the three classification techniques on WEKA toolkit. After conducting the experiments the Naïve Bayes classifier achieves the highest accuracy followed by the SVM classifier, and C4.5 classifier respectively. The SVM requires the lowest amount of time to build the model needed to classify Arabic documents, followed by Naïve Bayes Classifier, and C4.5 classifier respectively.