CRII:OAC:Towards Automated Ontology Generation for Software Evolution Utilizing Large Language Models

Software development processes produce large volumes of data that are often overlooked, difficult to visualize, and challenging to revisit or trace over time. These assets contain valuable knowledge that, if properly extracted, can support meaningful visualization and intelligent reasoning. Ontology is widely recognized as a practical approach to knowledge representation and enables efficient reasoning over structured information. A key challenge in ontology generation for software processes is that its automation cannot scale to handle the continuously accumulating and evolving data. This project aims to resolve this issue by innovatively applying large language models to automatically generate ontologies from data created in software engineering life cycles.

By bridging knowledge representation and software process, this project produces ontologies that enable intelligent reasoning for software development tasks and broader stakeholder engagement with software systems. The project introduces custom metrics to assess the usability of the ontologies for downstream tasks such as change impact analysis, version comparison, and code generation. It also develops tools for supporting the ontology generation automation workflow and visualizing and refining ontologies. The resulting ontology repositories and software tools contribute to improved understanding and automation in evolving software systems.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

Award Number: 2451397
Principal Investigator: Songhui Yue
Funds Obligated: $174,408
State: SC