본문 바로가기
KERD
논문KCI 등재

대형언어모델(LLM) 소프트웨어 사용 품질 특성의 중요도 연구

A Study on the Importance of Software Quality-in-Use Characteristics for Large Language Models

0

조회

0

다운로드

0

인용

0

참고문헌

키워드

주제분류공학일반

초록

Abstract : Large Language Models (LLMs) have become central to search, document creation, and customer service tools, but existing research has predominantly focused on accuracy or acceptance, leaving software quality in use insufficiently addressed. To address this gap, this study applied the ISO/IEC 25010 framework and Analytic Hierarchy Process (AHP) to determine the importance weights of 5 main characteristics and 10 sub-characteristics. Fourteen LLM experts participated in pairwise comparisons, with all consistency ratios (CR) meeting the 0.10 threshold. The analysis revealed that Freedom from Risk (0.30) had higher priority than Satisfaction (0.29) and Context Coverage (0.21), while Effectiveness and Efficiency were rated relatively lower. These findings suggest that future LLM service development should prioritize minimizing economic and informational risks, while enhancing trustworthy usefulness and multi-environment adaptability.

Abstract

Abstract : Large Language Models (LLMs) have become central to search, document creation, and customer service tools, but existing research has predominantly focused on accuracy or acceptance, leaving software quality in use insufficiently addressed. To address this gap, this study applied the ISO/IEC 25010 framework and Analytic Hierarchy Process (AHP) to determine the importance weights of 5 main characteristics and 10 sub-characteristics. Fourteen LLM experts participated in pairwise comparisons, with all consistency ratios (CR) meeting the 0.10 threshold. The analysis revealed that Freedom from Risk (0.30) had higher priority than Satisfaction (0.29) and Context Coverage (0.21), while Effectiveness and Efficiency were rated relatively lower. These findings suggest that future LLM service development should prioritize minimizing economic and informational risks, while enhancing trustworthy usefulness and multi-environment adaptability.

관련 논문AI추천

이 간행물의 다른 논문 (4건)