nick@ZERMATT.LCS.MIT.EDU (Nick Papadakis) (08/27/88)
Date: Mon, 15 Aug 88 10:59 EDT From: finin@PRC.Unisys.COM To: ailist@ai.ai.mit.edu Subject: Software Reusability: An Intelligent Approach (UNISYS) Software Reusability: An Intelligent Approach Mark A. Simos and James Solderitsch Software Technology Department UNISYS Paoli Research Center GVL-2 Auditorium Unisys Great Valley Labs 12:00-1:00, 15 August 1988, The topic of software reusability has been at the forefront of software engineering research for quite some time, but as yet has failed to live up to initial expectations. Part of the reason for this failure was early and lingering confusion about the subjects of software engineering and software reusablity, and the belief that the proper software engineering methodology, and perhaps even the right programming language, would naturally and effortlessly lead to the creation of reusable software. Recent research has begun to pinpoint the unique issues relating explicitly to software reusability. This talk describes a practical approach to software reuse based on the incremental development of intelligent libraries of reusable components. Such libraries, or repositories, are structured around explicit domain models which are knowledgebased frameworks providing taxonomic representations of specific application domains. These frameworks provide a uniform view of both static software components and generative capabilities, and contain tools to actively guide users in browsing among and selecting existing components, or classifying and qualifying new candidate components for the repository. After an introduction to some of the essential issues of software reusability, we present some background motivation for a domain-specific focus to reusability. We next discuss the use of program generation and knowledge-based techniques that support domain-specificity and sketch the evolution of the development of a reuse library based on these techniques. We close with a description of our current project that is directed at developing the basic Reusability Library Framework (RLF) technology necessary for the development of such domainspecific libraries. The RLF project is sponsored by the STARS Ada Foundations Technology program (contract number N00014-88-C-2052). Specific objectives of the RLF project include providing a set of knowledge-based components in Ada that support the creation and maintenance of domain models, and the development, using this platform, of general library tools for component testing, qualification and retrieval.