Date of Graduation
This work describes and presents a process for building a neural network-based reuse economic model. This study adopts a bottom-up perspective where granularity moves from the developer (software program) level to the enterprise level. The goal of the proposed process is to create a model which provides accurate and reliable results which can be applied to various forms of software reuse in various programming paradigms. The current state-of-practice for measuring effort and reuse effort is extended, leading to improvements in accuracy. Furthermore, in order to accomplish this goal, it becomes necessary to address different forms of reuse: black-box and white-box reuse and different perspectives of reuse: design-for-reuse and design-with-reuse. The reuse economic model must operate for both procedural and object-oriented programming paradigms. A more realistic definition of what constitutes reuse is described. This work expands upon the previous works by the author including: [Boetticher 93a] which verifies and validates the use of neural networks for assessing software; [Boetticher 93b] which produces a neural network-based reusability metric; and [Boetticher 95] which describes a process of using neural networks to determine â€œhot spotsâ€ in a software repository.
Boetticher, Gary Dawson, "A neural network-based bottom-up approach for building a software reuse economic model." (1999). Graduate Theses, Dissertations, and Problem Reports. 8503.