Characterising LLM-Generated Competency Questions: a Cross-Domain Empirical Study using Open and Closed Models
For ontology engineers, this work provides a systematic framework to evaluate and select LLMs for automated CQ generation, but the findings are incremental as they confirm known variability in LLM outputs without introducing new methods or benchmarks.
This paper introduces quantitative measures to characterize LLM-generated Competency Questions (CQs) for ontology engineering, analyzing readability, relevance, and structural complexity across open and closed models. Results show distinct generation profiles shaped by the use case, with no single model dominating across all dimensions.
Competency Questions (CQs) are a cornerstone of requirement elicitation in ontology engineering. CQs represent requirements as a set of natural language questions that an ontology should satisfy; they are traditionally modelled by ontology engineers together with domain experts as part of a human-centred, manual elicitation process. The use of Generative AI automates CQ creation at scale, therefore democratising the process of generation, widening stakeholder engagement, and ultimately broadening access to ontology engineering. However, given the large and heterogeneous landscape of LLMs, varying in dimensions such as parameter scale, task and domain specialisation, and accessibility, it is crucial to characterise and understand the intrinsic, observable properties of the CQs they produce (e.g., readability, structural complexity) through a systematic, cross-domain analysis. This paper introduces a set of quantitative measures for the systematic comparison of CQs across multiple dimensions. Using CQs generated from well defined use cases and scenarios, we identify their salient properties, including readability, relevance with respect to the input text and structural complexity of the generated questions. We conduct our experiments over a set of use cases and requirements using a range of LLMs, including both open (KimiK2-1T, LLama3.1-8B, LLama3.2-3B) and closed models (Gemini 2.5 Pro, GPT 4.1). Our analysis demonstrates that LLM performance reflects distinct generation profiles shaped by the use case.