论文部分内容阅读
四、不确定性推理由于数据和专门的知识都常常是不确定的,所以对现实世界的问题直接使用以上的演绎推理方法非常复杂。因此,专家系统的设计者们不得不放弃要求系统具有逻辑完备性,而采用有效的启发式方法,以利用那些难免出错但却有效的判断性知识,这些知识是人类专家对特定问题探索出来的。现在我们来比较一下已用于不确定性推理的一些方法。 4.1 似然推理假设A是一个关于客观事物的断言,诸如一个三元组(属性,对象,值),那么人怎样处理同这个断言有关的不确定性呢?把不确定性数量化的经典形式是概率论,但已提出并使用了其他的一些理论。其中有确定性理论,可能性理论及Dempster/shafer的证据理论。我们依次来讨论这四种理论,但着重前两种理论。
IV. Uncertainty reasoning Because data and specialized knowledge are often uncertain, it is very complicated to use the above deductive reasoning directly on the real-world problems. As a result, designers of expert systems have had to abandon the system that requires the system to be logically complete, using effective heuristics to exploit the inevitable but valid judgmental knowledge that human experts have explored for specific problems . Now let’s compare some of the methods that have been used for uncertainty reasoning. 4.1 Likelihood Inference Assuming A is an assertion of objective things, such as a triple (attributes, objects, values), how does one deal with the uncertainty associated with this assertion? The classical form of quantifying uncertainty Probability theory, but some other theories have been proposed and used. Among them are the theory of certainty, the theory of possibility, and Dempster / Shafer’s theory of evidence. We discuss these four theories in turn, but focus on the first two theories.