Degree

Doctor of Philosophy (PhD)

Department

Computer Science

Document Type

Dissertation

Abstract

Software changes are performed frequently to allow cutting edge features to be delivered to users in a timely manner and to assuage the ever increasing security vulnerabilities found in software. These changes, while necessary, can leave users dissatisfied with software that they use. Security patches can slow down computer performance, and extensive user interface changes can leave users dissatisfied. This research presents a methodology for identifying which software changes have high potential to cause user dissatisfaction. As a part of the methodology, we use natural language processing to analyze the software or system requirements business rule text to determine which changes, if they are implemented, will likely most prominently affect users. We identify these changes by analyzing the change text for keywords that indicate the change has potential to affect the user interface or user experience (UI/UX). We analyze both the proposed changes document and the original or current requirements document, called the baseline. We define a connectedness metric that is based on changes and related requirements and is used to quantify where a change is likely to impact the current state of the software system. We further define transitive relationships which are derived from the change to baseline mappings, baseline to baseline mappings, and change to change mappings. These different types of relationships help detect multiple levels of dependencies among requirements and changes. We use these relationships to quantify the potential impact of the proposed change on UI/UX. We also investigate whether the impact of UI/UX changes differs from the impact of other changes. We use 17 datasets in our experiments. The methodology can aid requirements writers in consolidating disjointed requirements and can help link changes to requirements in the original or current requirements document. It can also help detect which software features are most likely to cause user dissatisfaction.

Date

11-2-2021

Committee Chair

Carver, Doris L.

DOI

10.31390/gradschool_dissertations.5685

Download

Share

COinS