Satisficing in knowledge-based systems

摘要

A satisficing knowledge-based system reaches a satisfactory resolution as quickly as possible. A satisfactory resolution is one that reaches an answer that may not be optimal, but is satisfactory or heuristically “close enough”. This work focuses on the principles that can be used to support a methodology that will produce satisficing decisions in a data-driven knowledge-based system. A codified satisficing methodology that has been implemented is also presented. The purpose of satisficing is to resolve as few initial condition and goal states as can be managed consistent with reaching a satisfactory answer. This is within the normal constraints of an expert system, namely that the graph resolution would appear to be “natural” to a human user. This requires that the inference engine satisfactorily exhaust a particular topic before moving onto the next topic. Satisficing constraints, when attained, indicate that there is sufficient supporting information for a rule and that the evidence for it does not need to be further considered.