Subjective Difficulty Estimation for Interactive Learning by Sensing Vibration Sound on Desk Panel

In this paper, we propose a method which estimates the student's subjective difficulty with a vibration sound on a desk obtained by a microphone on the back of the desk panel. First, it classifies the student's behavior into writing and non-writing by analyzing the obtained sound data. Next, the subjective difficulty is estimated based on an assumption that the duration of non-writing behavior becomes long if the student feels difficult because he (or she) would not have progress on answer sheet. As a result, the accuracy of the proposed so simple behavior classification reaches around 80%, and that of the subjective difficulty estimation is 60%.