By analyzing and study Apriori algorithm, this paper finds two main problem in the practical use: Firstly generating quite a lot of redundancy association rules; Secondly, it maybe mine an association rule that possess high support and confidence, but it’s uninteresting and even it is false. Furthermore, it can’t generate a rule with negative item. Thus, this paper bring two properties of the association rules and introduce the third interesting measure’s threshold ---correlation support measure,by use of the properties, it can diminish some redundancy, at the same time we can use the correlation support measure to mine those rules which are more interesting and it design a efficiency algorithm and by the use of induce two properties to raise efficiency in the algorithm to resolve the above shortcoming well.