Privacy preserving DBSCAN algorithm for clustering

Abstract

In this paper we address the issue of privacy preserving clustering. Specially, we consider a scenario in which two parties owning confidential databases wish to run a clustering algorithm on the union of their databases, without revealing any unnecessary information. This problem is a specific example of secure multi-party computation and as such, can be solved using known generic protocols. However there are several clustering algorithms are available. They are applicable to specific type of data, but DBSCAN [4] is applicable for all types of data and the clusters obtained by DBSCAN are similar to natural clusters. However, DBSCAN [4] algorithm is basically designed as an algorithm working on a single database. In this paper we proposed a protocols for how the distances are measured between data points, when the data is distributed across two parties. By using these protocols we propose the first novel method for running DBSCAN algorithm operating over vertically and horizontally partitioned data sets, distributed in two different databases in a privacy preserving manner. © Springer-Verlag Berlin Heidelberg 2007.