Blackboard systems at the architecture level

摘要

Since Hearsay-II, many application systems using the blackboard architecture have been built. Some of these applications have been built using expert system shells or tools. These tools contain the skeletal constructs of blackboard systems into which the user adds the necessary information and knowledge from the application domain to create an application system. Although the tools are useful in the implementation phase, they provide little help in the design of the system, which must necessarily precede the implementation phase. The determination of the blackboard levels, the knowledge sources, and the details of the control strategy is a task reserved for the knowledge engineer/system designer. In order to automate the design process, one needs to understand the relationships between the architectural constructs and domain information that provide the semantics of the architecture. In this paper the authors explore the components of blackboard systems from the perspective of the type of knowledge that goes into the makeup of blackboard systems at the architecture level. They describe a research system, KASE, that captures the various knowledge needed for design and applies the knowledge to aid knowledge engineers design blackboard systems.