KBRA: A new paradigm for requirements engineering

The authors present a knowledge-based requirements assistant (KBRA) that is a component of the knowledge-based software assistant (KBSA). The idea behind KBSA is to create a knowledge-based life-cycle paradigm spanning software development from requirements to code and to formalize software practice so that computers can be used as active reasoning agents in developing software. The authors identify knowledge-representation issues associated with requirements acquisition and analysis, and note the three realms in which mechanisms operate to resolve knowledge issues: presentations, structured text, and evolving system description. They describe artificial intelligence techniques used to provide consistent reasoning processes for the intelligent assistant: inheritance of properties from generic object types, automatic classification based on discriminators indicating how to specialize instances, and constraint propagation for processing ramifications of requirements decisions and for supporting retraction when people change their minds.<>