Insights Into Student Attention During Online Lectures: A Classification Approach Using Eye Data.

With the growing prevalence of online learning, ensuring student engagement and attention during online lectures has become a challenge. Unlike traditional classroom settings, where students are physically present and can interact with their peers and the instructor, online lectures can feel more passive and isolating. To address this challenge, we investigate the use of eye-tracking data to identify distractions during online lectures. By analyzing this data, it becomes possible to gain insights into a person's attention and focus. To achieve this, we employed a technique to approximate the raw gaze data using piecewise linear functions, where each segment represents an eye movement event such as fixation and saccade. These segments are then used to extract important features that distinguish between eye gaze time series before and after the distractor stimuli in the online lecture. We then train a binary classifier using the extracted features and also rank the importance of the features. The classifier achieves an accuracy of 73.6 % in classifying gaze timeseries as a distraction or no distraction.