A Conceptual Framework for Real-Time Emotional-State Monitoring of Students in VLEs to Identify Students at Risk

Abstract

Virtual Learning Environments (VLEs) feature rich textual data which lend itself naturally to the identification and monitoring of aspects of students’ interactions. While reducing attrition and improving performance remain the primary objectives of learning analytics, we contend that student contributed text can be used to learn about emotions and other extra-rational features. This would help provide a response to the recent cries for help from the sector, seeking a system looking to address the worrying mental health crisis trends. This paper addresses these issues by discussing the necessary mechanisms within a conceptual framework which would sit in a VLE and capture emotional state changes in the students’ interaction style or tone. For such a framework, the aim would be to help educators to carry out timely interventions when a potential cause of distress is identified. Experimental results on available datasets from education and psychology serve as a feasibility study for these tasks, and offer a perspective on the potential of the approach.