Specific Domain Ontology Construction Using Large Language Models

This article presents the use of Large Language Models (LLMs), specifically GPT-3.5 and GPT-4, to automate the construction of ontologies for the Brazilian maritime territory. The study involved creating twenty ontologies that were evaluated by human experts, revealing that while the models generated coherent conceptual hierarchies, the outputs required refinement to fully represent the domain context. This research highlights the potential of LLMs in ontology development, addressing the challenge of manual crafting in specific domains.