Content area
Software reuse refers to the use of previously developed software resources in new applications. In this dissertation, we concentrate on the concept of widespread reusability which means (1) reuse by other software developers within the same organization as well as the original developer, (2) reuse of data and process objects produced during the systems analysis and design phases as well as code, (3) reuse of general and specific purpose software resources across a variety of application domains, and (4) reuse of software resources along a continuum of task types from maintaining existing systems to developing new software systems.
The main objective of this dissertation is to search for and experiment with tools and techniques to support widespread software reusability. A survey of the literature was carried-out and it was concluded that achieving widespread software reusability would require a total systems approach and the development of specialized tools and techniques. This dissertation suggests extending the concept of CASE software towards a "Reusability Support System" (RSS) that provides automated support for the classification, storage, retrieval, and modification of software resources of all kinds.
A major impediment to software reusability is the difficulty of classifying various software resources in such a way that another person can easily identify them when they are needed to satisfy some new requirements. An important requirement for widespread reusability therefore is an effective means for classifying and retrieving software resources from all phases of software system development.
To address this problem, three issues are investigated in this research: how to represent various reusable software resources for both classification and retrieval, how to organize a library of reusable software resources, and how to support the specification of user requirements.
A frame-based knowledge representation scheme employing concepts of abstraction is proposed to represent reusable software resources from all phases of the life cycle. A classification scheme based on the faceted method from library science is proposed. A software resource network, which is integrated into the faceted method, is used for organizing software resources. To provide more effective descriptive information, a "software resource model" (SRM) is proposed which emphasizes the explicit capture of semantic information as a means of increasing the precision of the classification- retrieval process. To demonstrate the feasibility and functionality of the proposed SRM approach, a prototype system was developed. In addition, an experiment testing the SRM approach against a keyword approach was carried-out using student subjects. The experiment suggests that the SRM approach can help in achieving more consistent classifications; however, the students found the approach quite difficult to understand and use. Improvements to the SRM model that may overcome these problems are suggested in the conclusion to the dissertation.