Detection of learning styles with prior knowledge data using the SVM, K-NN and Naïve Bayes algorithms

Abstract

The two types of automatic learning style detection approaches are data driven (DD) and literature based (LB). Both methods of automatic learning style detection have advantages over traditional learning style detection methods because they use external data sources, such as forums, quizzes and views of teaching materials, that are more accurate than the questionnaires used in traditional styles of detection. The results of automatic detection, on the other hand, do not always reflect learning styles. This paper presents a learning style recognition method that uses data from the learner’s internal source, namely prior knowledge, to overcome these challenges. Prior knowledge is proposed because it is based on the learner’s knowledge or skills, which better reflect the learner’s characteristics, rather than on the learner’s behaviour, which tends to be dynamic. By using past knowledge, this paper presents a method for detecting automatic learning patterns. The learning style detection framework is unique in that it consists of three stages: prior knowledge question development, prior knowledge measurement and learning style detection using the Support Vector Machine (SVM), Naïve Bayes and K-Nearest Neighbour (K-NN) classification methods. The accuracy of learning style detection using prior knowledge data was higher than detection results using behavioural data or hybrid data (prior knowledge + behaviour) in this study