Generative AI Decision Support Tool for Automated Traceability During Systems Development Integrating Large Language Models (LLMs) and Retrieval Augmented Generation (RAG)

The US Department of Defense reportedly spends four percent of life cycle costs on requirements traceability (Powers & Stubbs, 1985) and GlobalData estimated that approximately $17.5 billion was spent on digital thread efforts in 2023 (GlobalData, 2023). Further, many organizations, including the US Food and Drug Administration, International Electrotechnical Commission (IEC), and the Capability Maturity Model Integration (CMMI), now require bi-directional requirements traceability (US FDA, 2023), (US Government, 2022), (IEC 61508), (Vassilka Kirova, 2008).Current methods for establishing and managing traceability during systems development are ineffective, prohibitively expensive, and do not scale for the future needs of managing complex systems inundated with hundreds of thousands of interrelated data artifacts across the engineering life cycle. This praxis aims to empower systems engineers to tackle the requirements traceability problem by delivering a decision support tool for establishing relationships automatically leveraging the latest GPT and LLM capabilities available. This praxis demonstrates the tool effectively reduces the level of effort and time while maintaining a higher level of quality compared to existing methods.

Author

• Baker, Jason

Language

• English

Keyword

• Large Language Model (LLM)
• Knowledge Graph
• Retrieval Augmented Generation (RAG)
• Generative AI

Date created

• 2024

Type of Work

• Dissertation

License

• All rights reserved

Rights statement

• In Copyright

GW Unit

• Systems Engineering

Degree

• D.Engr.

Advisor

• Fong, Titus

Committee Member(s)

• Blackford, J.P.
• Fong, Titus
• Etemadi, Amir

Persistent URL

•  https://scholarspace.library.gwu.edu/etd/00000076b