Date of Award
8-2013
Document Type
Thesis
Degree Name
Master of Science (MS)
Legacy Department
Mechanical Engineering
Committee Chair/Advisor
Mocko, Gregory M
Committee Member
Summers , Joshua Summers D
Committee Member
Fadel , Georges
Abstract
The objective of this research is to use computational linguistics to identify semantic implicit relationships between text-based relationships. Specifically, natural language processing is used to implement linguistic semantics in requirement analyzers. Linguistic semantics is defined as the meaning of words beyond their string form, part of speech, and syntactic function. Many existing design tools use part of speech tagging and sentence parsing as the foundation of their requirement analysis but ultimately use string algorithms to evaluate requirements. These string algorithms cannot capture the implicit knowledge in requirements. This research compares five methods of requirement analysis. A manual analysis provides the benchmark against which the subsequent analyzers are judged. A syntactic analysis is implemented and compared to the manual method to gain insight into the capabilities of current methods. The other three analyzers implement semantic tools for requirement analysis through semantic ontologies and latent semantic analyses. The results from the semantic analyzers are compared to the results of the other two analyzers to judge the capabilities of semantics in requirement analysis. The findings show that semantics can be identified with at least 74% accuracy. Further, the agreement between the semantic results and the manual results are more related than the syntax results and the manual results. While the implementation of semantics into requirement analysis does not completely agree with manual findings, the semantic analyses improve upon syntactic and string matching analyses used in current research.
Recommended Citation
Lash, Alex, "COMPUTATIONAL REPRESENTATION OF LINGUISTIC SEMANTICS FOR REQUIREMENT ANALYSIS IN ENGINEERING DESIGN" (2013). All Theses. 1698.
https://open.clemson.edu/all_theses/1698