Further Theoretical Contributions to a Privacy Preserving Distributed OLAP Framework

This paper complements our privacy preserving distributed OLAP framework proposed in [8] by introducing four major theoretical properties that extend models and algorithms presented in [8], where the experimental validation of the framework has also been reported. Particularly, the framework [8] makes use of the CUR matrix decomposition technique [12] as the elementary component for computing privacy preserving two-dimensional OLAP views effectively and efficiently. Here, we investigate theoretical properties of the CUR decomposition method, and identify some theoretical extensions of this method, which, according to our vision, may result in benefits for a wide spectrum of aspects in the context of privacy preserving distributed OLAP, such as privacy preserving knowledge fruition schemes and query optimization.