Towards Ontology-Guided Requirements Quality Analysis: an LLM-Based Tool for EHR Systems

Abstract

Systematic assessment of software requirements against standardized quality models remains a challenge in requirements engineering (RE), particularly in large-scale, critical, and regulated information systems. Large language models (LLMs) have been extensively studied in many RE tasks; however, grounding LLM-assisted requirements quality analysis in formal quality frameworks remains underexplored. This paper introduces a three-stage Requirements Quality Analysis (RQA) workflow that identifies relationships between requirements, performs ontology-guided analysis to make quality-related issues and missing information explicit, and implements this workflow in the RQA tool for structured analysis and human review. The approach uses a predefined concept set based on the Ontology of quality model concepts presented in ISO/IEC 25002:2024 as a grounding structure for LLM-assisted analysis. The tool is demonstrated with real requirements from the Finnish Apotti electronic health record (EHR) system as an illustrative case of a critical and regulated healthcare software context. The results illustrate how ontology-guided LLM analysis can support more structured and traceable identification of quality-related gaps and review points in requirements.