A proposed perspective shift: viewing specification design as a planning problem

Abstract

Our interest is in formal models of the specification process. In this paper we argue that with the right perspective and set of assumptions, work in AI planning may be brought to bear on three aspects of designing a software specification: proposing a specification (generation); analyzing a proposed specification for problems (critiquing); and modifying a specification to mitigate problems (elaboration). We describe a program called ASAP which automates portions of the specification process.