@article{ART003228306},
author={Tae-Ho Min},
title={Design and Evaluation of LLM-Based Conversational Agents for Diagnosing Science Learning Difficulties},
journal={Journal of The Korea Society of Computer and Information},
issn={1598-849X},
year={2025},
volume={30},
number={7},
pages={87-96}
TY - JOUR
AU - Tae-Ho Min
TI - Design and Evaluation of LLM-Based Conversational Agents for Diagnosing Science Learning Difficulties
JO - Journal of The Korea Society of Computer and Information
PY - 2025
VL - 30
IS - 7
PB - The Korean Society Of Computer And Information
SP - 87
EP - 96
SN - 1598-849X
AB - This study explores the use of large language model-based conversational agents to diagnose science learning difficulties. Four prototype models were developed by varying the classification schemes of science learning difficulties and the inclusion of emotional support, and simulation-based experiments were conducted by using ChatGPT-4o in the roles of students. The results showed that models using detailed classifications of environmental factors achieved higher diagnostic accuracy than those using detailed classifications of individual factors, while the inclusion of emotional support did not have a statistically significant effect. Based on these findings, an improved model was designed and applied to twelve students in the ninth grade. Analysis of the interactions revealed that certain areas, including community-related factors, were sometimes omitted from the diagnosis. In addition, the diagnostic process was prematurely terminated when students deviated from the topic or refused to continue the conversation. Drawing on these results, the study proposes design recommendations for conversational agents aimed at diagnosing science learning difficulties.
KW - Large language model;Generative AI;Conversational agent;Prompt engineering;Science education;Learner diagnosis
DO -
UR -
ER -
Tae-Ho Min. (2025). Design and Evaluation of LLM-Based Conversational Agents for Diagnosing Science Learning Difficulties. Journal of The Korea Society of Computer and Information, 30(7), 87-96.
Tae-Ho Min. 2025, "Design and Evaluation of LLM-Based Conversational Agents for Diagnosing Science Learning Difficulties", Journal of The Korea Society of Computer and Information, vol.30, no.7 pp.87-96.
Tae-Ho Min "Design and Evaluation of LLM-Based Conversational Agents for Diagnosing Science Learning Difficulties" Journal of The Korea Society of Computer and Information 30.7 pp.87-96 (2025) : 87.
Tae-Ho Min. Design and Evaluation of LLM-Based Conversational Agents for Diagnosing Science Learning Difficulties. 2025; 30(7), 87-96.
Tae-Ho Min. "Design and Evaluation of LLM-Based Conversational Agents for Diagnosing Science Learning Difficulties" Journal of The Korea Society of Computer and Information 30, no.7 (2025) : 87-96.
Tae-Ho Min. Design and Evaluation of LLM-Based Conversational Agents for Diagnosing Science Learning Difficulties. Journal of The Korea Society of Computer and Information, 30(7), 87-96.
Tae-Ho Min. Design and Evaluation of LLM-Based Conversational Agents for Diagnosing Science Learning Difficulties. Journal of The Korea Society of Computer and Information. 2025; 30(7) 87-96.
Tae-Ho Min. Design and Evaluation of LLM-Based Conversational Agents for Diagnosing Science Learning Difficulties. 2025; 30(7), 87-96.
Tae-Ho Min. "Design and Evaluation of LLM-Based Conversational Agents for Diagnosing Science Learning Difficulties" Journal of The Korea Society of Computer and Information 30, no.7 (2025) : 87-96.
