Large Language Models for Automatic Standardization of Cyber Deception Plans based on the Adversary Engagement Ontology

Adversary Engagement Ontology (AEO) is a candidate ontology for the Unified Cyber Ontology (UCO), a community effort aimed at ontological standardization of cyber domain concepts and objects under a unifying framework. It forms a part of the Cyber Domain Ontology (CDO). In the past, community efforts and development have always been labor-intensive with regards to changes in ontology, example generation for adopters, and documentation generation. Large Language Models (LLMs), such as Claude-3.5-Sonnet and GPT4, have been proven capable of automating many tasks and aiding in human expert decision-making. Additionally, LLMs have been used in code interpretation, generation, and evaluation with efficiency and accuracy comparable to that of humans. This emergent capability of LLMs has led to the advantage of using LLMs to streamline the process of ontology development. Motivated by the aforementioned-approaches, we aim to demonstrate how these foundational LLMs can assist in ontology example generation and development, as well as be utilized to automate structured, albeit tedious tasks.