Abstract: This paper presents a new model, "discrimination tree" and its "parasitic tree", to more efficiently study expert systems. The discrimination tree is a tree to systematically describe discriminating knowledge. The parasitic tree is a discrimination tree used to determine the sub-features of the nodes of another discrimination tree. The trees can be used to systematically describe complicated discriminating knowledge and to clearly design high-performance classification inference engines. In addition. They can be easily translated into a knowledge base in the form of production rules or frames.