Dissertation Information for Gregory W. Hislop NAME: - Gregory W. Hislop
- (Alias) Greg Hislop
DEGREE:
- Ph.D.
DISCIPLINE:
- Library and Information Science
SCHOOL:
- Drexel University (USA) (1993)
ADVISORS: - Gary Wayne Strong
COMMITTEE MEMBERS: - Stephen J. Andriole - Daniel Joyce-Villanova - Richard Harold Lytle - James Solderitsch
MPACT Status: Fully Complete
Title: Assessing the potential for software reuse
Abstract: The idea of software reuse is to build software from components that can be applied repeatedly. This approach may provide a significant improvement in the productivity of software work. It may reduce costs of both development and maintenance.
On the other hand, a formal reuse strategy requires financial commitment and organizational change. In addition, a variety of technical and organizational factors must be present for reuse to succeed. This indicates that an organization needs an initial assessment of reuse potential before implementation.
This project explores 3 questions related to making this preliminary assessment. In that context, it emphasizes techniques that are relatively inexpensive and that use an organization's existing software portfolio. The questions are: (1) Is a simple, single term functional classification adequate to describe existing software? (2) Can analysis of software form locate similar (reused) programs in a software portfolio? (3) Is there any discernable relationship between software form and function?
The project includes case studies examining these questions for 3 large commercial organizations. Various aspects of the case settings vary substantially, but results are quite consistent across all 3 cases.
The study data indicate that a fairly simple functional classification scheme is adequate for an initial assessment. However the most basic schemes previously proposed are probably not sufficient.
The analysis of software form shows good potential for locating similar and reused modules in existing software. This analysis applies pattern metrics developed to detect plagiarism in student programs to locating reused modules in commercial software. It also applies common software metrics, singly and in combination, to the task of locating reused modules.
This type of analysis could help provide an assessment of the overall level of similarity in the existing portfolio. It could also be useful in identifying initial candidates for reusable components.
Finally, the data do not reveal any interesting relationship between software form and function.
|