Towards conceptually-oriented software requirements analysis and design.

Abstract

Software developers currently do not have sufficient machine support for knowledge management. This thesis describes a partial solution to some of these knowledge management problems in software development, emphasizing conceptually-oriented requirements analysis and design, which we believe is fundamental to the software development process. We propose an ontological framework which consists of a domain-independent ontology, a methodology ontology, and an application ontology. This framework permits multiple methodologies and applications to be integrated. The domain-independent ontology provides knowledge common to all domains, while the methodology ontology assists the designer to capture essential design knowledge following a specific methodology. The top level of the application ontology provides common structure over multiple applications. Lower levels of the application ontology correspond to each specific application, consistent with the domain-independent ontology and methodology ontology. This ontological framework is supported by a generic knowledge management system known as CODE, which manages all knowledge in a knowledge base. A telecommunication example is developed to illustrate the approach. ObjecTime, an object-oriented commercial design CASE tool, is used to derive an executable design. Thus, we demonstrate that a knowledge management system (KMS) like CODE can be a powerful tool to provide intelligent machine support for conceptually-oriented analysis and design.