Large Language Models (LLMs) have been used to automate ontology verification, serving as an effective substitute for human experts. This study explores the potential of LLMs as cognitive agents for refining OntoClean-validated hierarchies by adhering to OntoClean meta-properties and criterion validation based on OntoClean principles. The study automates the extraction of specific identity and unity criteria for concepts and validates whether child nodes in a hierarchy inherit and align with the criteria of their parent nodes. The evaluation conducted on FoodOn and AEON demonstrates the ability of the LLM-based mechanism in applying established ontological principles effectively. It offers a practical alternative to manual ontology verification, ensuring transparency in the process.