GazeMotif: A Gaze Speed Motif based Learning Behaviour Analysis Framework for Video Learning

In this paper, we propose a gaze-based analysis framework for video learning, named GazeMotif, which aims to analyze the characteristics of eye movement in various teaching areas during video learning, thereby providing a novel perspective for understanding eye movement behavior. By collecting gaze speed data and processing it through visibility graph algorithms and sliding window methods, we constructed and labeled motifs, developed motif transition graphs, and performed R/S analysis. We performed an experiment to evaluate GazeMotif, and the results indicated that this analysis framework can reveal patterns of gaze speed during video learning.