Model based software engineering using situation theory with applications to business services

Abstract

Natural language specifications tend to be incomplete, imprecise and mutually inconsistent. Most software development methodologies rely on stakeholder reviews as a feedback mechanism. Generally, stakeholders have business domain expertise, but lack the technical background required to comprehend the detailed, low-level formalisms commonly used. As a result, the specifications for review may either retain much of the informal, natural language aspects and thus lead to incorrect decisions during subsequent analysis and refinement stages, or detailed specifications in a formal notation are presented and under-informed approvals are granted in spite of difficulties in understanding the formalised specifications. In either case, the resulting business system implementations are riddled with errors, delivered late and over budget. Model based software engineering approaches have been successfully applied in technical programming. Although many business software development methodologies have been introduced, widespread success with them has been elusive. The search for a theory that bridges from informal natural language text to a mathematical logic notation lead to the consideration of situation theory as a suitable candidate. Situation theory emerged from research combining logic and linguistics. Unlike precise formalisms, such as first-order logic and set theory, situation theory admits information that is incomplete and/or inconsistent. The studies in situation theory include the analysis of informal language and its transformation into formal language, that is, potentially providing means of bridging from informal to formal models. In this thesis we investigate how situation theory could be applied to the various stages of model based software engineering methods and facilitate the refinement of natural language requirements documents through steps to fully formal models.