Hybrid Cloud-Based Privacy Preserving Clustering as Service for Enterprise Big Data

Abstract

Clustering as service is being offered by many cloud service providers. It helps enterprises to learn hidden patterns and learn knowledge from large, big data generated by enterprises. Though it brings lot of value to enterprises, it also exposes the data to various security and privacy threats. Privacy preserving clustering is being proposed a solution to address this problem. But the privacy preserving clustering as outsourced service model involves too much overhead on querying user, lacks adaptivity to incremental data and involves frequent interaction between service provider and the querying user. There is also a lack of personalization to clustering by the querying user. This work “Locality Sensitive Hashing for Transformed Dataset (LSHTD)” proposes a hybrid cloud-based clustering as service model for streaming data that address the problems in the existing model such as privacy preserving k-means clustering outsourcing under multiple keys (PPCOM) and secure nearest neighbor clustering (SNNC) models, The solution combines hybrid cloud, LSHTD clustering algorithm as outsourced service model. Through experiments, the proposed solution is able is found to reduce the computation cost by 23% and communication cost by 6% and able to provide better clustering accuracy with ARI greater than 4.59% compared to existing works.