Theses and Dissertations

Date of Award

8-1-2024

Document Type

Thesis

Degree Name

Master of Science in Engineering (MSE)

Department

Mechanical Engineering

First Advisor

Satya Aditya Akundi

Second Advisor

Constantine Tarawneh

Third Advisor

Hiram Moya

Abstract

This thesis explores the applications of natural language processing (NLP) techniques in model-based system engineering (MBSE) to help generate System Modeling Language (SysML) diagrams. MBSE is a method that aids in enhancing traditional engineering practices by modeling to help improve understanding and communication in systems development. SysML, one of the modeling languages for MBSE, helps represent a system's architecture, behavior, and information flow. Translating systems requirements and specifications into SysML models can be time-consuming and can lead to errors when created manually. Automating the creation of SysML diagrams from textual descriptions with the help of NLP techniques can aid in faster realization and reduce modeling errors. This will simplify the initial stages of system design, ensuring precision and uniformity in models. Despite the promise, generating SysML diagrams using NLP faces challenges against natural language’s inherent complexity and the need for significant domain knowledge, leading to challenges in extracting and interpreting system requirements from natural language text. This thesis reviews three commonly used NLP-based approaches to generate system representation from natural language text, i.e., Rule-based, Machine Learning (ML)-based, and Hybrid methods. The rule-based method relies on predefined rules to map text to SysML elements, the Machine Learning method learns from data to identify relationships and patterns, and the Hybrid method aims to combine the strengths of both. Further, a Rule-based framework is proposed to partially automate the generation of SysML diagrams to address some of the challenges identified. The framework demonstrates its effectiveness in creating a SysML representation by implementing an application for generating Class diagrams. The proposed framework underlines the potential issues related to natural language variability and complexity, paving the way for a more streamlined generation of system architectural representations.

Comments

Copyright 2024 Joshua Andre Ontiveros. https://proquest.com/docview/3115714498

Download

Share

COinS