Considerations for making blackboard shells more general in order to improve the knowledge engineering process are described. An architecture for general blackboard shells is then proposed. It includes a set of generalized functions for constructing and supporting various operations on different types of blackboards, i.e. planning, constraint, domain, and context blackboards. A general structure for knowledge sources is included to facilitate versatile categories of knowledge representations. Various control strategies are developed for the inference engine that will allow users to selectively configure specific inference engines for specific problems.
展开▼