Development of a meta-blackboard shell

Abstract

The architecture of a meta-blackboard shell is described, which can be specialized as specific blackboard shells to fit specific domains. The knowledge source structure of the shell is designed as a miniature blackboard architecture to increase the local inference capability through the generalization of knowledge representation. The control component is designed as a restricted blackboard architecture to propose a generalized focus-of-attention control strategy that searches for a set of best-matched pairs of knowledge sources and solution areas as the next focus under dynamic constraints. Various problem-solving paradigms are integrated through planning and constraint blackboards. Several high-order mechanisms, such as viewpoint-based reasoning are supported. The blackboard structuring system is designed to support both homogeneous and heterogeneous decompositions of a problem-solving space.<>