A logical approach to requirements analysis
The increasingly competitive nature of the market for advanced computer-based systems has led to a demand for automated tools, which can help improve the quality of systems engineering products, and at the same time reduce their cost and the time it takes to develop them. This problem has been investigated by selecting a logic programming language to represent both a customer's requirements and the rules of the methodology used to analyze them. Part of a typical requirements document was manually symbolized into Prolog ready to be processed by the rules of structured analysis. Symbolization was found to be essentially equivalent to the parsing and clarification that normally have to be done at the beginning of any project. The rules of analysis were constructed by investigating exactly what else the engineer does to create a graphical model from the customer's document. A paradigm for requirements analysis was created to show how the Prolog-implemented rules can be applied in an iterative fashion through interaction with the analyst. The initial result of applying the rules to the requirements was compared with the final data flow diagrams obtained during manual analysis, and proved to be a valid basis for system development, lacking only those facts known to the analyst but omitted from the document. It was also demonstrated how the paradigm can be extended to other analysis methodologies, to later phases of a development project, and to the traceability of requirements. Thus it appears that a logic programming language is an effective way to represent clear requirements, to express the rules of requirements analysis, and even to represent the design database throughout a development lifecycle. The automation that it facilitates has good potential for cost-saving and for improving the responsiveness of a product to its requirements.
Systems design|Computer science|Aerospace materials
Scott, Peter Crosby, "A logical approach to requirements analysis" (1993). Dissertations available from ProQuest. AAI9321472.