Development of a generalized knowledge source structure for blackboard systems

摘要

A generalized knowledge source structure is proposed for blackboard systems so that knowledge sources can play a more active role in the reasoning process. The structure is designed as a miniature blackboard architecture containing a set of knowledge source elements, a micro-blackboard, and a control knowledge source element. A knowledge source builder is implemented in a user-friendly, graphic environment so as to realize a structure with powerful knowledge representation schemes and effective inference abilities. The system represents a knowledge source at two levels: the epistemological level and the heuristic level. At the heuristic level, it represents the microblackboard structure in an extended frame system, and provides many heuristic functions to speed up the reasoning process. To reduce the loading of experts, the epistemological level is developed on the basis of an augmented semantic net theory, so that the experts can directly express domain knowledge without knowledge of the internal structure of a knowledge source. The system automatically translates the knowledge at the epistemological level into the internal structure of the heuristic level, which is ready for efficient reasoning. This approach renders the incremental building of knowledge sources easy and natural.