Analysison PrivacyPreservationin Vertically Partitioned Datain Cloud Using Homomorphic Encryptionand Rule Mining Algorithms

Abstract

Privacy has become decisive in knowledge-driven applications in the distributed outsourced data. In cloud, data are fragmented as horizontal and vertical partition and it is very challenging for the data owners to protect their private data from unauthorized users in the outsourced database. To ascertain security, the cryptography technique is adopted with public key secured document in the cloud. In this paper, a privacy-preserving architecture is designed for vertically portioned data in cloud with the help of Homomorphic encryption. Two popular data analysis techniques, Frequent Itemset Mining and Association Rule Mining are adopted to build the Association Rule among the Frequent Items in an encrypted transaction of vertically partitioned database. In this work, rule mining generation algorithms such as Eclat, Apriori and FP-Growth are taken for analysis in terms of computation time and scalability of data.  The analysis result shows that eclat algorithm is less time consuming to generate rule in the cloud, irrespective of number of transactions.

Keywords: Frequent itemset, Encryption, Eclat, Apriori, FP-growth, Association Rule Mining