MOOClm: Learner Modelling for MOOCs

Abstract

Massively Open Online Learning systems, or MOOCs, generate enormous quantities of learning data. Analysis of this data has considerable potential benefits for learners, educators, teaching administrators and educational researchers. How to realise this potential is still an open question. This thesis explores use of such data to create a rich Open Learner Model (OLM). The OLM is designed to take account of the restrictions and goals of lifelong learner model usage. Towards this end, we structure the learner model around a standard curriculum-based ontology. Since such a learner model may be very large, we integrate a visualisation based on a highly scalable circular treemap representation. The visualisation allows the student to either drill down further into increasingly detailed views of the learner model, or filter the model down to a smaller, selected subset. We introduce the notion of a set of Reference learner models, such as an ideal student, a typical student, or a selected set of learning objectives within the curriculum. Introducing these provides a foundation for a learner to make a meaningful evaluation of their own model by comparing against a reference model. To validate the work, we created MOOClm to implement this framework, then used this in the context of a Small Private Online Course (SPOC) run at the University of Sydney. We also report a qualitative usability study to gain insights into the ways a learner can make use of the OLM. Our contribution is the design and validation of MOOClm, a framework that harnesses MOOC data to create a learner model with an OLM interface for student and educator usage.