Abstract :
The rule in rule-based system can be divided into two categories: inference rule and knowledge expressed by ldquoif..., then...rdquo Therefore, rule-based system can be divided into inference rule system and hypothetical inference system. In inference rule system, the general law is regarded as the inference rule, used to infer the unknown facts or theorems from the known ones. This inference system is concise, needs skills. It is suitable for human. But, there are many inference rules, a computer cannot choose from them automatically. So, it is not suitable for computer. A computer is suitable to make inference that is repetitive, mechanical, does not need skills. In hypothetical inference system, the general law is expressed by ldquoifldrldrldr, thenldrldrldrrdquo and regarded as the theorem and taken as the major premise, hypothetical inference is regarded as the sole inference rule. Because there is only one inference rule, every time a computer wants to make inference, the computer employs it. As to the numerous known facts or theorems, the computer searches them top-down, from left to right, depth first plus backtracking, as the way Prolog does. This inference system is not concise, not need skill, but is repetitive, mechanical. So, hypothetical inference system is not suitable for human but is suitable for computer. Inference rule system can be transformed into hypothetical inference system.
