Privacy preserving data mining survey of classifications

Abstract

With the rapid progress of data mining, protecting sensitive information is a critical issue before sharing to outside parties. Association rules become a powerful tool in discovering hidden sensitive information among parties, whereas some of them try to protect their own sensitive information by blocking inference channels. For instance, two parties are selling same products in different markets. After sharing their sale information to each other, one tries to analyze the adversary sale information (that we call it data mining). So, after analyzing and accessing to adversary’s beneficial information, the party changes the arrangement of his products in his market since it helps the customers to access products more easily and quickly. However the rate of selling in his market will increase, the other party is analyzing the information which has been censored (that what we call it privacy preserving in data mining). The methods that the parties has been trying to hide before sharing sensitive information and the characteristics of the algorithms have been discussed in this paper.