A Conceptual Framework for Monitoring the Emotional State of Students in VLEs

Abstract

The number of higher education students facing mental health issues has reached a critical point, with major consequences both in terms of academic achievement and general health and wellbeing. Institutions recognise the issue and have put in place a number of measures to try and counteract the crisis. Monitoring students’ wellbeing can play a critical role, and Virtual Learning Environments could be instrumental in achieving this goal. This is particularly true in online-learning scenarios, where face-to-face contact is less present or not present at all, and educators need to rely on observing the online behaviour. However, the information overload to lecturers is significant, keeping track of each online discussion forum is extremely onerous, and the help from "learning analytics" not always useful, as they often concentrate on measures of students’ performance, engagement, and presence in the virtual classroom, which are not necessarily good indicators of mental health. The work presented in this thesis proposes to bridge this gap by addressing the ef- fectiveness of emotion or writing style profiling to identify students at risk. The work proposes a conceptual framework for a system, intended to sit alongside the virtual learning environment, and able to play the role of an "emotion observer", identifying and flagging potential issues to the educator. We propose a system where technology is supportive of educators rather than replacing them, and is not intrusive or changing the classroom dynamics. We demonstrate the validity of the approach with a series of experiments. We address the technical feasibility of such a system by investigating how established artificial intelligence techniques, and "off-the-shelf" tools implementing them, can be used to carry out the tasks that would need to be performed by a system implementing the approach, and we discuss their performances on either available datasets, or, for one of the experiments, a purpose built dataset, which is part of the contributions of the thesis. We address the admissibility of such an approach by conducting a focus group study with a group of experts in online learning. The contribution of the thesis is therefore the first complete feasibility study on the de- velopment of a novel system able to monitor students’ emotional state, both individually and as a cohort, and longitudinally over the course of their studies, which is aimed at supporting online educators identify students at risk and implement strategies for intervention.