Abstract: Association rule is one of the important modes in data mining, and Apriori algorithm is a typical algorithm of mining association rule. However, it has disadvantages: globally searching the database and deleting significative rules by minimum support threshold during creating large itemset. The Maradbcm algorithm is a new algorithm of mining association rule, which is proposed based on the KDD’s inner mechanism, and it can overcome the Apriori algorithm’s disadvantage. After briefly describing the double-base cooperating mechanism and Maradbcm algorithm, the algorithm applies to mushroom database, and the result indicates that the algorithm is effective. This fully shows the important influence of inner mechanism on the KDD mainstream development, and provides a new path for the research of the whole knowledge discovery system.