A Modular Cybersecurity Ontology Constructing Method Driven by Large Language Models

Cybersecurity ontologies form the foundation for intelligent systems in security domains. Current high-quality cybersecurity ontologies require labor-intensive manual collaboration between ontology engineers and security experts. To improve construction efficiency, this study developed a large language model-assisted modular ontology learning method for cybersecurity. The approach represents modular ontologies through multiway forest structures and optimizes three key construction tasks via large language models: concept identification, domain relevance determination, and triple extraction. Experimental results demonstrate the method's capability to semi-automatically construct high-quality modular security ontologies.