Constructing and analyzing specifications of real world systems
Kaizhi Yue · 1986
The specification of real world systems is essential in developing application software and building AI expert systems. However, in formalizing our intuitions about systems, omissions and confusions cause errors. Correctly describing the behavior of a system can be a formidable task. In this dissertation we propose that the semantic basis of representation languages lies in the mappings from observable situations to abstractions. In particular, we present such a mapping that serves as a framework for representing a broad class of real world systems and for analyzing and detecting errors in their specifications. At the basis of this framework are the causal and teleological interrelations that entail constraints among system categories. Specifications are analyzed against these constraints. A language that incorporates these system categories has been designed. A symbolic simulator, which plays a similar role to an interpreter for a programming language, has been developed for this language. A comprehensive specification environment, DAO, has been implemented that can allow the user to edit, browse, and type-check as well as to simulate and analyze specifications.