This paper describes a knowledge-based approach to automate CODARTS, a software design method for concurrent and real-time systems. The approach uses multiple paradigms to represent knowledge embedded within CODARTS. Semantic data modeling provides the means to model specifications and related designs. A specification meta-model enables automated inferences about the presence of semantic concepts within a specification, while a design meta-model permits automated reasoning concerning concurrent designs. Production rules form the basis for modeling a set of heuristics that can generate concurrent designs based upon semantic concepts from the specification and design meta-models. Together, the semantic data models and production rules, encoded using an expert-system shell, compose CODA, an automated designer's assistant. CODA is applied to generate ten concurrent designs for four real-time problems.
