Towards Class-level Lifecycle Governance of Ontology-driven Conceptual Models for Software Quality Assurance

Abstract

Context. Ontology-driven conceptual models evolve over time as domain modellers introduce class-level extensions to express taxonomies, constraints, and domain-specific knowledge. Problem. Monitoring the quality of these conceptual models during evolution is challenging. Existing approaches in the literature mainly address version-centric evolution, overlooking governance of individual class-level lifecycles. Consequently, domain modellers lack support for specifying lifecycle policies to monitor class-level evolving taxonomies over time. In addition, the lack of policies demands the implementation of tests manually for each class, which is time-consuming, error-prone, and difficult to scale. Method. In this paper, we propose OCMLG (Ontology-driven Conceptual Modelling Lifecycle Governance), a model-driven development tool for monitoring the quality of ontology-driven conceptual models based on lifecycle policies. OCMLG allows domain modellers to automatically generate tests based on a specific lifecycle policy and monitor the conceptual models’ class-level lifecycle quality over time. Results. We present a preliminary evaluation in a simulated environment where LLM-based agents evolve an ontology-driven conceptual model across multiple iterations. OCMLG successfully detected breaking class-level lifecycle changes and reported the conceptual model’s quality metrics. Conclusions. Our results suggest that OCMLG provides a structured and automated mechanism for governing the class-level lifecycle of ontology-driven conceptual models. We expect OCMLG to integrate with the AI-SQUARE platform ecosystem to maintain and evolve software quality assurance metrics.