A robust sampling-based framework for privacy preserving OLAP

Abstract

A robust sampling-based framework for privacy preserving OLAP is introduced and experimentally assessed in this paper. The most distinctive characteristic of the proposed framework consists in adopting an innovative privacy OLAP notion, which deals with the problem of preserving the privacy of OLAP aggregations rather than the one of data cube cells, like in conventional perturbation-based privacy preserving OLAP techniques. This results in a greater theoretical soundness, and lower computational overheads due to processing massive-in-size data cubes. Also, the performance of our privacy preserving OLAP technique is compared with the one of the method Zero-Sum, the state-of-the-art privacy preserving OLAP perturbation-based technique, under several perspectives of analysis. The derived experimental results confirm to us the benefits deriving from adopting our proposed framework for the goal of preserving the privacy of OLAP data cubes. © 2008 Springer-Verlag Berlin Heidelberg.