Abstract: The research of classification methods is the focus in the field of business intelligence,also is widely used in information retrieval,ecommerce and information resource management.In recent years,the association rule mining method expands to classification areas,and shows some good features.This paper focuses on a novel correlation classification method (that is,GARC class method based on the information entropy), summarizes the basic idea,the main properties and methods.This method not only keeps the tranditional association classification method in the comprehensible,efficiency and precision,and reflects other aspects,and effectively eliminates the possibility of redundant rules and conflict.In addition,it carries on the corresponding fuzzy extensions (GARCf) according to the “sharp boundary” problem of numerical attribute discrete division.

Key words: Information entropy, Class associative rule, Review