Transforming Brainwave Signals into Symbolic Strings Towards Academic Emotion Recognition

Abstract

Students tend to experience varying academic emotions while engaged in learning activities. It is important for affective systems to predict certain emotions, particularly negative emotions, to be able to provide proper remediation and improve the learning experience of students. In this study, patterns of brainwave signals or electroencephalogram (EEG) of academic achiever high school students, such as those formed before the onset of an academic emotion, are analyzed by transforming this data into symbolic strings using a modified Shapelet Transform and SAX and from these strings, determining the emotional state of the students. These strings are shortened to further ease analysis. Results have shown that strings transformed from the same EEG feature and same emotion are significantly different to strings from different features and different emotions.