Identifying patterns of epistemic emotions with respect to interactions in massive online open courses using deep learning and social network analysis

Convincing evidence found by educators and psychologists shows that learners' interactions in discussion forums in massive online open courses (MOOC) overwhelmingly affect their epistemic emotions. In a MOOC context, epistemic emotions, such as the experiences of curiosity, enjoyment, confusion, and anxiety, are caused by the cognitive equilibrium or incongruity between new information and existing knowledge while learning via a MOOC course. Therefore, uncovering the relationships among epistemic emotions and interactions from largescale MOOC data is an important task. By gathering multiple data generated by 1190 Chinese learners, this study employed a combination method of deep learning and social network analysis (SNA) to identify patterns of epistemic emotions with respect to interactions on a MOOC platform. The results revealed that four patterns, identified from core, neighbor, scattered, and peripheral learners, tended to expand relationships by votes and construct deep communication by comment and reply interactions. Of particular interest, the core and neighbor learners' patterns demonstrated significantly higher interactions and epistemic emotions than the scattered and peripheral learners' patterns.