Toward New Method for Adaptive Learning

Introduction

The majority of research on adaptive education systems focused on modeling of learners based on their levels of knowledge and preferences (Bull et al. 2007). In parallel, several experimental studies in differential psychology have shown that individual differences play an important role in teaching (Zhang, 2015). Learners acquire and process pedagogical information in different ways; they learn at different rates depending on the learning context (Sternberg, 2015). These results suggest that adapting pedagogical sequences to learning styles allows the learner to retain information longer and to apply it more effectively.

The goal of this research is to describe a suitable learner model, consisting of the cognitive ability and pedagogical preference of a learner that has associated cognitive metrics found within instructional content. Our approach consists to split the learner model into a knowledge model and a preferences model. We try to propose a solution inspired by the statistical theory of learning for the implementation of the model: Bayesian network.

The learner model maintains knowledge and preferences. The decision engine uses the knowledge model and performance obtained by the learner. It determinates the following pedagogical objective to be taught. It also predicted the theoretical performance that the learner can achieve if pedagogical sequence is optimally adapted. It also uses the preference model. It reinforces the estimation of the learning style and inferences the metadata of the ideal sequence to be proposed to the learner. These metadata are then provided to the adaptation module which uses the content model to select the optimal pedagogical sequence and adapt the content.