An adaptive recommender-system based framework for personalised teaching and learning on e-learning platforms

Abstract

Current learning management systems such as Moodle and WebCT are considered as linear systems that provide e-learning material in a fixed-sequence, delivering the same content to learners regardless of their differences in background knowledge. For learners engaged in self-study online distance learning, this may result in material being presented at either too high or too low cognitive levels. According to the Zone of Proximal Development (ZPD) theory, this may result in either frustration or boredom among learners. This paper proposes a recommender-system-based adaptive e-learning framework for personalised teaching on e-learning platforms. The framework would assist designers, teachers and learners to identify issues they need to consider in order to address challenges of poor engagement in online distance settings, arising from a “one-size-fits-all” approach that does not recognise the role of individual differences in teaching and learning. Secondly, the framework may enable the identification of problems or obstacles that may be encountered when supporting learners in their quest to reduce frustration and boredom when using a Recommender-Based Pedagogical System (RBPS). A literature review was conducted on adaptive e-learning systems based on the ZPD theory, learner modelling, the Generic Adaptive Framework and a recommendation system model. 70 articles were selected from a database of 720 articles published between 2010 and 2017 to come up with the dimensions needed to develop such a model for the framework through deductive analysis. The research found out that the majority of the studies only consider three dimensions to an adaptive framework, that is, the learner model, the content model and the adaptation engine while the Generic Adaptation Framework proposes seven dimensions. In addition, the majority of the studies are based on the principles of macro-adaptation which provide a “static” snapshot of a learner's profile instead of dynamically adjusting the adaptation as learner variables. In the proposed adaptive framework, we identified five dimensions, including real-time dynamic adaptation and context modelling in addition to the learner model, the domain model and the pedagogical strategy.