An Empirical Approach to the Utilization of Affective Decision Tree Models in Smart Teaching

Abstract

In the upcoming age of AI, smart education reshapes not only the macro modality of society and education but also the micro modality of knowledge and learning. The top priority of smart education is the timely learning disability diagnosis and feedback, the assessment and improvement of students' mental quality, in which affective data is the crucial index. Putting in perspective the working of positive and negative affects in smart teaching, this research proposes the affective decision tree model drawing on the affective data collected from the PANAS test. Based on the C5.0 algorithm, classification rules are extracted from the calculation of the information gain ratio of affective variables and the construction of decision tree is realized through SPSS 26.0. The research findings indicate that 1) negative affects took precedence over positive affects in the formulation of participants' smart learning strategy; 2) the architecture of the affective decision tree constructed via learners' affective judgments and ratings represents the affective filtering process which optimizes teachers' selection and organization of smart teaching and learning activities. Accordingly, the affective decision tree model could be used as the efficient prediction model and affective profiling model and teaching support system in smart teaching, prompting the affective and cognitive connectivity between teachers and students.