Dissertation/Thesis Abstract

Metrics for Aspect Mining Visualization
by Jorgensen, Gisle J., Ph.D., Nova Southeastern University, 2018, 85; 10842982
Abstract (Summary)

Aspect oriented programming has over the last decade become the subject of intense research within the domain of software engineering. Aspect mining, which is concerned with identification of cross cutting concerns in legacy software, is an important part of this domain. Aspect refactoring takes the identified cross cutting concerns and converts these into new software constructs called aspects. Software that have been transformed using this process becomes more modularized and easier to comprehend and maintain. The first attempts at mining for aspects were dominated by manual searching and parsing through source code using simple tools. More sophisticated techniques have since emerged including evaluation of execution traces, code clone detection, program slicing, dynamic analysis, and use of various clustering techniques. The focus of most studies has been to maximize aspect mining performance measured by various metrics including those of aspect mining precision and recall. Other metrics have been developed and used to compare the various aspect mining techniques with each other. Aspect mining automation and presentation of aspect mining results has received less attention. Automation of aspect mining and presentation of results conducive to aspect refactoring is important if this research is going to be helpful to software developers. This research showed that aspect mining can be automated. A tool was developed which performed automated aspect mining and visualization of identified cross cutting concerns. This research took a different approach to aspect mining than most aspect mining research by recognizing that many different categories of cross cutting concerns exist and by taking this into account in the mining process. Many different aspect mining techniques have been developed over time, some of which are complementary. This study was different than most aspect mining research in that multiple complementary aspect mining algorithms was used in the aspect mining and visualization process.

Indexing (document details)
Advisor: Mitropoulos, Francisco J.
Commitee: Mukherjee, Sumitra, Sun, Junping
School: Nova Southeastern University
Department: Computer Information Systems
School Location: United States -- Florida
Source: DAI-B 79/12(E), Dissertation Abstracts International
Subjects: Computer Engineering, Computer science
Keywords: Aspect, Metrics, Mining, Visualization
Publication Number: 10842982
ISBN: 978-0-438-20858-2
Copyright © 2020 ProQuest LLC. All rights reserved. Terms and Conditions Privacy Policy Cookie Policy