The Privacy Preservation of Data Cubes

Abstract

As the development of OLAP, users are not only inner analysts or managers but also customers, partners or even third parties. This makes the privacy preservation more and more important. Based on the observation that a single data item in a data cube is not likely to be accessed alone, but a number of data are often aggregated to give summaried information and the trends of database, in this thesis, we proposed two simple but effective methods to distort sensitive individual data of data cube but the summation of range query was almost remained the same as that of the original data. The experiments were done on APB benchmark data set from OLAP council. The results showed that our methods achieved better privacy preservation and better accuracy for range queries than traditional random data distortion alternatives.