Presentation Information
[A-11-08]Analysis of the Relationship between Questions to a Generative AI Chatbot and Learner Understanding in Programming Education
◎Ayumi Kawashima1, Mana Morikawa1, Takahiro Baba1 (1. Kurume Institute of Technology)
Keywords:
Generative AI,Programming Education,Learning Analytics,Understanding Estimation,Chatbot
Generative AI-based educational support systems have recently gained widespread attention, and chatbot-assisted learning has become increasingly common in programming education. However, most existing systems do not consider learners’ levels of understanding and therefore provide limited personalized support. This study investigates the relationship between questions submitted to a generative AI chatbot and learner performance in a programming course. The analysis was conducted using 1,207 question logs collected from 106 students. Learners were divided into high- and low-understanding groups based on assignment scores. TF-IDF-based keyword extraction and question-type analysis were applied to identify differences between the groups. The results showed that high-understanding learners tended to ask questions related to program comprehension and feature enhancement, whereas low-understanding learners more frequently sought specific implementation methods and direct solutions. These findings suggest that learner understanding may be inferred from question content, providing a foundation for future adaptive learning support systems.