Supervised classification on navigational behaviours in web-based learning systems to identify learning styles

Abstract

Several education hypermedia systems (EHS) use learning styles (LS) as a criterion for adaptation and tracking. To measure these styles, EHS are generally based on the questionnaires provided by the used LS model, and that learners should answer before the first session. This approach has a major drawback: learners' LS are defined only once. To overcome this limitation, recent researches are currently being done on the detection of LS based on learner's interaction traces. Their general criticism is related to the use of a specific environment, and therefore specific traces and indicators. Therefore, we aim to identify the learner's LS automatically, based on simple navigation traces. In this paper, we present experimental results of identification of sequential/global and active/reflective LS, for 45 students, using supervised classification. The findings provide initial evidence that LS can be automatically identified based on learners' navigation behaviours.