UNDERSTANDING AFFECTIVE DYNAMICS OF LEARNING TOWARD A UBIQUITOUS LEARNING SYSTEM

Abstract

Understanding student learning behaviors is of prime importance for educational research. Many complex factors influence learning processes, but one collective impact of all these factors is how they affect learning and the degree of motivation. In this study, we discuss the current state of human affect detection in education, our proposed affect change model and its implications. This study adopts dataset from ASSISTments online learning platform, which consists of student interaction data, and ground truth labels for affect states coded by Baker Rodrigo Ocumpaugh Monitoring Protocol (BROMP) certified coders to develop and validate the affect change model. We show that the proposed affect change model in combination with the adoption of machine learning algorithms will support the development of a ubiquitous learning system that tracks the student learning process within the context of contributing factors and provide interventions when needed.