@article{ART003280361},
author={Changgou Kang and Namgyu Kim},
title={A Study on the Application and Quality Comparison of LLM-Based Task-Oriented Dialogue Data Generation Methodology in the Korean Cafe Domain},
journal={Journal of The Korea Society of Computer and Information},
issn={1598-849X},
year={2025},
volume={30},
number={12},
pages={61-72}
TY - JOUR
AU - Changgou Kang
AU - Namgyu Kim
TI - A Study on the Application and Quality Comparison of LLM-Based Task-Oriented Dialogue Data Generation Methodology in the Korean Cafe Domain
JO - Journal of The Korea Society of Computer and Information
PY - 2025
VL - 30
IS - 12
PB - The Korean Society Of Computer And Information
SP - 61
EP - 72
SN - 1598-849X
AB - HR-MultiWOZ methodology has recently emerged as an efficient approach for generating Task-Oriented Dialogue (TOD) data with Large Language Models (LLMs). It introduced the first synthetic TOD dataset in the Human Resources (HR) domain, highlighting cost-effectiveness through schema-based design and minimal human input. However, the influence of prompt design on data quality and its applicability to non-English languages or domains beyond HR remain underexplored.
Accordingly, this study is conducted as a case study to empirically examine the applicability of the HR-MultiWOZ methodology to Korean and café ordering. Results show that simple prompt adjustments can effectively control the characteristics of LLM-generated dialogue (LGD), underscoring the methodology’s scalability and the pivotal role of prompt design in shaping dialogue data quality.
KW - Task-Oriented Dialogue;Synthetic Data Generation;LLM;HR-MultiWOZ;Prompt
DO -
UR -
ER -
Changgou Kang and Namgyu Kim. (2025). A Study on the Application and Quality Comparison of LLM-Based Task-Oriented Dialogue Data Generation Methodology in the Korean Cafe Domain. Journal of The Korea Society of Computer and Information, 30(12), 61-72.
Changgou Kang and Namgyu Kim. 2025, "A Study on the Application and Quality Comparison of LLM-Based Task-Oriented Dialogue Data Generation Methodology in the Korean Cafe Domain", Journal of The Korea Society of Computer and Information, vol.30, no.12 pp.61-72.
Changgou Kang, Namgyu Kim "A Study on the Application and Quality Comparison of LLM-Based Task-Oriented Dialogue Data Generation Methodology in the Korean Cafe Domain" Journal of The Korea Society of Computer and Information 30.12 pp.61-72 (2025) : 61.
Changgou Kang, Namgyu Kim. A Study on the Application and Quality Comparison of LLM-Based Task-Oriented Dialogue Data Generation Methodology in the Korean Cafe Domain. 2025; 30(12), 61-72.
Changgou Kang and Namgyu Kim. "A Study on the Application and Quality Comparison of LLM-Based Task-Oriented Dialogue Data Generation Methodology in the Korean Cafe Domain" Journal of The Korea Society of Computer and Information 30, no.12 (2025) : 61-72.
Changgou Kang; Namgyu Kim. A Study on the Application and Quality Comparison of LLM-Based Task-Oriented Dialogue Data Generation Methodology in the Korean Cafe Domain. Journal of The Korea Society of Computer and Information, 30(12), 61-72.
Changgou Kang; Namgyu Kim. A Study on the Application and Quality Comparison of LLM-Based Task-Oriented Dialogue Data Generation Methodology in the Korean Cafe Domain. Journal of The Korea Society of Computer and Information. 2025; 30(12) 61-72.
Changgou Kang, Namgyu Kim. A Study on the Application and Quality Comparison of LLM-Based Task-Oriented Dialogue Data Generation Methodology in the Korean Cafe Domain. 2025; 30(12), 61-72.
Changgou Kang and Namgyu Kim. "A Study on the Application and Quality Comparison of LLM-Based Task-Oriented Dialogue Data Generation Methodology in the Korean Cafe Domain" Journal of The Korea Society of Computer and Information 30, no.12 (2025) : 61-72.
