A novel approach for efficient mining and hiding of sensitive association rule

Abstract

Data mining is the process of analyzing large database to find useful patterns. The term pattern refers to the items which are frequently occurring in set of transaction. The frequent patterns are used to find association between sets of item. The efficiency of mining association rules and confidentiality of association rule is becoming one of important area of knowledge discovery in database. This paper is organized into two sections. In first part of paper an Improved Apriori algorithm is being presented that efficiently generates association rules. These reduces unnecessary database scan at time of forming frequent large itemsets. In second part of this paper we have tried to give contribution to improved apriori algorithm by hiding sensitive association rules which are generated by applying improved Apriori algorithm on supermarket database. In this paper we have used novel approach that strategically modifies few transactions in transaction database to decrease support and confidence of sensitive rule without producing any side effects. Thus in the paper we have efficiently generated frequent itemset sets by applying Improved Apriori algorithm and generated association rules by applying minimum support and minimum confidence and then we went one step further to identify sensitive rules and tried to hide them without any side effects to maintain integrity of data without generating spurious rules.