Rule-based detection of emotions in the Khan Academy platform

Abstract

The current relevance of Massive Open Online Courses (MOOCs) has provoked researchers in educational technology to work towards improving their pedagogical outcomes. Adaptive MOOCs are an example within this context. Given the importance of affective information within the adaptive systems, we propose a set of models to detect four emotions known to correlate with learning gains. The implementation of the models and the initial results from its application in a case study dataset are also provided.