Question Classification Based on Cognitive Levels using Linear SVC

Abstract

A student’s cognitive level can be determined through an assessment such as final examination. A person needs to have skills and knowledge with regard to educational assessments to make sure the questions are concurrent with the cognitive level. The aim of this paper is to find the best classifier to classify exam questions based on cognitive levels. The experiment is conducted in two phases. The first phase is to find the best mapping for SVM classifier (One-Versus-One and One-Versus-All). The classifier that produces the best result for mapping is used in the second phase for Naïve Bayes, KNN and Linear SVC. The result showsthat Linear SVC with OVO is the best classifier with 74.8% for f-measure and tf-idf as feature extraction which really benefits in increasing the classifier’s result. In future, the classifier will be tested to classify questions in the Malay language