Expert systems represent knowledge in sets of ‘if-then-else’ rules. Propositions are proven by discovering a chain of rules. This may be done through either forward or backward chaining. In forward chaining. Expert systems have had a measure of success on certain well-defined probllsms. There have, however, been many
failures. Most of the problems have arisen from the difficulty of obtaining a consistent, complete set of rules. The rule acquisition process requires the interaction of two specialists: a knowledge engineer and a domain expert. There are several problems with this process. First, the domain experts may not be fully co-operative. After all, they are being asked to participate in the development of a
system that will replace them with a computer. More often, domain experts only appear to be unco-operative; they simply cannot create a set of rules that explain how certain things are done. From one point of view, only trivial things can be reduced to a set of rules;all important aspects of human intelligence are beyond a simple rule set because a human expert’s knowledge is often in the form of natural language with imprecision or vague information.
对不起,是翻译成中文。。