Arabic Questions Classification Using Modified TF-IDF

Classifying the cognitive levels of assessment questions according to Bloom’s taxonomy can help instructors design effective assessments that are well aligned with intended learning outcomes. However, classification process is time consuming and requires experience. Many studies have attempted automate by utilizing different machine text mining approaches, but none has examined Arabic questions. The purpose this study examine research gap introduce a new feature extraction method would better suit their unique characteristics. It also aims provide Arab tool them automatically classify To accomplish purpose, developed dataset more than 600 then proposed modified term frequency-inverse document frequency (TF-IDF) for extracting features from Unlike traditional TF-IDF, was designed take nature into consideration. evaluated comparing it two methods been used classifying English questions, i.e., TF-IDF called part-of-speech-inverse (TFPOS-IDF). A t-test utilized whether difference in performance between three statistically significant. outperformed other methods. overall accuracy, precision, recall scored were significantly higher those TFPOS-IDF evaluation results indicate promising potential method, which be extended languages.