REU • Research Experience for Undergraduates
From Concepts to Competency
(2024)
Developing Skill Ontologies with LLM Assistance
![Group 75.png ha](https://res.cloudinary.com/cearto/image/upload/c_fit,h_800,w_1600/v1722016831/Group_75.png_ha.png)
Description
In recent decades, ontologies have become valuable tools for sharing and understanding domain-specific knowledge. Large Language Models (LLMs) have simplified the process of creating these ontologies by assisting in generating competency questions (CQs), enumerating concepts, and forming class hierarchies. However, their application to skill ontologies remains underexplored. This paper explores using LLMs to develop skill ontologies in digital fabrication, specifically laser cutting. We propose an iterative method for developing a laser cutting skill ontology, consisting of three phases: ideation, refinement, and validation. In the ideation phase, we generate CQs, concepts, and skills using various prompting techniques. During refinement, we link skills and form hierarchies. In validation, we evaluate the ontology’s accuracy. To assess our approach, we conduct card sorting interviews with digital fabrication experts. By leveraging LLMs for skill ontology development, this work aims to enhance skill acquisition and foster educational tool development across various domains, improving access to quality education
![Group 75.png ha](https://res.cloudinary.com/cearto/image/upload/c_fill,g_face,h_100,w_100/v1722016831/Group_75.png_ha.png)