Privacy Preserving Optics Clustering

Abstract

OPTICS is a well-known density-based clustering algorithm which uses DBSCAN theme without producing a clustering of a data set openly, but as a substitute, it creates an augmented ordering of that particular database which represents its density-based clustering structure. This resulted cluster-ordering comprises information which is similar to the density based clustering’s conforming to a wide range of parameter settings. The same algorithm can be applied in the field of privacy-preserving data mining, where extracting the useful information from data which is distributed over a network requires preservation of privacy of individuals’ information. The problem of getting the clusters of a distributed database is considered as an example of this algorithm, where two parties want to know their cluster numbers on combined database without revealing one party information to other party. This issue can be seen as a particular example of secure multi-party computation and such sort of issues can be solved with the assistance of proposed protocols in our work along with some standard protocols.