Privacy-awareness of distributed data clustering algorithms revisited

Several privacy measures have been proposed in the privacy preserving data mining literature. However, privacy measures either assume centralized data source or that no insider is going to try to infer some information. This paper presents distributed privacy measures that take into account collusion attacks and point level breaches for distributed data clustering. An analysis of representative distributed data clustering algorithms show that collusion is an important source of privacy issues and that the analyzed algorithms exhibit different vulnerabilities to collusion groups.