Democratizing Emotion Research in Learning Sciences

We describe the design of an AIED tool EmoInfer to accelerate process-based research on emotions in Learning Sciences and help educational stakeholders understand the interplay of cognition and affect in ecologically-valid learning situations. Through an iterative implementation pipeline, we have developed a user interface to streamline automatic annotation and analysis of videos with facial expressions of emotion. EmoInfer can be applied to quantify and visualize the frequency and the temporal dynamics of naturally occurring or induced emotions (both on-the-fly and posthoc after data collection). By offering an accessible toolkit with "low floor, high ceiling and wide walls", we aim to initiate democratizing emotion research in Learning Sciences.