Security in Outsourcing of Association Rule Mining

Speaker: WONG Wai Kit

Abstract

Outsourcing association rule mining to an outside service provider can reduce cost. By using data encryption, data owner can prevent the service provider from accessing private information in the data. A study on adopting substitution cipher to encrypt transaction data for outsourcing association rule mining has been carried out. We studied the correctness problem of using one-to-n item mappings in the encryption. It was discovered that a one-to-n item mapping is not more secure than a one-to-one item mapping. To strengthen the security of a one-to-n mapping, we developed a transaction data encryption technique. An algorithm to encrypt transactions based on this method was developed. A comprehensive cryptanalysis has been carried out - the result confirms that the proposed technique is highly secure even if the service provider were given the knowledge (support counts) of most large itemsets in the original data.