Knowledge-based systems application to reduce risk in software requirements

Abstract

The development of complex large scale software systems begins with the elicitation and explication of requirements definitions between users and software systems engineers. Errors at this phase may account for 35 to 75 percent of the errors that will occur in software development. The development of requirements is probably flawed through failure in the processes of critical thinking on the part of the user and the design team. This leads to volatility in requirements in the form of imprecision, multiplicity, and conflict in quality goals. There exists a substantial need to provide a means for verification and validation (V&V) of requirements, particularly those factors that relate system characteristics to functionality factors. Risk related to these factors must be resolved if the error rate is to be reduced. Functionality goals have been classified and relationships established, a conflict matrix has been prepared, and a rule-set to apply the inference model has been constructed. The results of this work, on development of a knowledge-based system to resolve questions of risk in software requirements, show that substantial improvement in risk reduction is feasible by application of a rule-based system to reduce requirements volatility.