Barrier-free affective communication in MOOC study by analyzing pupil diameter variation

A MOOC (Massive Open Online Course) study shortens the distance between students and educators, and surpasses time and space, but it also creates barriers of true emotion interaction in the study process. This research demonstrated the feasibility of using a pupil diameter variation as the indicator of implicit affection states for the scales of valence and arousal, aiming to find a way of revealing the students' true emotion. In the experiment, affective music clips were selected as stimuli, the participants' pupillary responses information and their valence-arousal labeling scores were used for emotion recognition modeling. Multilayer Perceptron, Kstar and SVM algorithms were validated in the experiment for model comparison. SVM achieved the best recognition rate in both valence and arousal affective dimensions, indicating that affective states could be recognized by pupil diameter variation. On the basis of the recognition model, an application named "EMOOC" was developed for communication improvement in the MOOC's study, which could visualize students' emotional states to the MOOC teacher as feedback, bridging the emotion gap due to the online communication barrier.