Development of a Concern-Response System Based on Characteristics Analysis of Ideal Teachers in Stories

Abstract

If the ideal image of a teacher, particularly the structure of dialogue between students and teachers, can be clarified, it could be applied to AI-based educational agents and support systems. In this study, we focused on the characteristics of ideal teachers in stories, particularly the types of concerns students have and how teachers respond to them. We analyzed the structural features of concerns and their resolution. Additionally, based on the extracted features from our analysis, we developed a chatbot system that responds to user-inputted concerns with corresponding responses from the dataset. The analysis targeted scenes where students were troubled and where their motivation was enhanced. To organize the collected text data, we first performed grouping and categorization using the KJ method, followed by basic statistical analysis. Subsequently, we quantified the collected text data and conducted factor analysis and co-occurrence analysis using the numerical data.