( r , QI)-Transform: Reversible Data Anonymity Based on Numeric Type of Data in Outsourced Database

Abstract

An outsource database is a database service provided by cloud computing companies. Using the outsource database can reduce the hardware and software’s cost and also get more efficient and reliable data processing capacity. However, the outsource database still has some challenges. If the service provider does not have sufficient confidence, there is the possibility of data leakage. The data may has user’s privacy, so data leakage may cause data privacy leak. Based on this factor, to protect the privacy of data in the outsource database becomes very important. In the past, scholars have proposed k-anonymity to protect data privacy in the database. It lets data become anonymous to avoid data privacy leak. But k-anonymity has some problems, it is irreversible, and easier to be attacked by homogeneity attack and background knowledge attack. Later on, scholars have proposed some studies to solve homogeneity attack and background knowledge attack. But their studies still cannot recover back to the original data. In this paper, we propose a data anonymity method. It can be reversible and also prevent those two attacks. Our study is based on the proposed r-transform. It can be used on the numeric type of attributes in the outsource database. In the experiment, we discussed the time required to anonymize and recover data. Furthermore, we investigated the defense against homogeneous attack and background knowledge attack. At the end, we summarized the proposed method and future researches.