Date of Award

2026-08-01

Degree Name

Master of Science

Department

Engineering

Advisor(s)

Amit J. Lopes

Abstract

Model-based systems engineering (MBSE) promises a single authoritative model in place of static documents, but the translation of requirements documents into that model remains manual, slow, and dependent on scarce modeling expertise. This thesis asks whether an AI-assisted workflow can automate that translation, and whether the tooling itself can be built by vibe coding, in which a practitioner directs an AI in natural language rather than writing code directly. The result is a plugin for Magic Systems of Systems Architect (MSoSA) 2022x that reads a PDF requirements document, extracts requirements through the Anthropic Claude API, and generates a SysML model containing requirement elements, design blocks, test cases, a requirements diagram, all six SysML requirement relationship stereotypes, and stub placeholders for external references.

A feasibility evaluation applied the plugin to four synthetic documents spanning a quality spectrum from unstructured prose to a purpose-built specification with full traceability. On the three structured documents the plugin performed essentially perfectly: 100% extraction precision and recall, 100% link recall and correct typing across 237 declared relationships, zero alterations to requirement text, and import of the largest document in 81 seconds. On the prose document recall fell to 52%, and imported text often arrived reworded, merged, or split. The central finding is that document structure outweighs wording quality: every failure mode occurred on prose. AI assisted import is practical today on structured input, shifting the engineer's effort from transcription to review. The plugin itself, vibe coded by a systems engineer rather than a software developer, shows that engineering organizations can extend their own modeling tools as needed.

Language

en

Provenance

Received from ProQuest

File Size

93 p.

File Format

application/pdf

Rights Holder

Joshua Larson

Download

Included in

Engineering Commons

Share

COinS