Frequent Pattern Miningbased On Fully Homomorphic Encryption In The Cloud Environment

The big data have attracted more and more attention with the advent of the cloud era, people found that important information is often hidden behind a lot of huge data. Data mining is a kind of deep data analysis method, by which we can find all kinds useful information. But in the cloud environment, personal privacy and information security in data mining deserve great concern because misuse or abuse of data mining may lead to disclosure of users’sensitive information. How to do data mining while ensuring data security under the cloud environment has been a great issue that is not well-addressed. The recent breakthrough on fully homomorphic encryption (FHE) has been considered as a potential solution. FHE is a technique that can directly perform various operations on encrypted data without affecting their confidentiality.In this paper, we firstly reviews data mining, homomorphic encryption, and the integration of these two technologies. Secondly, we analyze the development of cryptography with a focus on FHE. We then design the main function modules of the FHE over integers, and implement the modules in the Linux environment. Thirdly, we discuss the association rules and the frequent pattern mining, in particular the Apriori algorithm of frequent pattern mining, and propose a binary Apriori algorithm that can be integrated with the FHE. Then, we design a variety of schemes that integrate the homomorphic encryption technique and frequent pattern mining model under the cloud environment. After a careful comparison of various schemes, we finally propose two algorithms for mining frequent patterns based on fully homomorphic encryption under the cloud environment, one of which is a low-traffic algorithm and the other is a low-computational-cost algorithm.This paper is the first one that integrates fully homomorphic encryption with frequent pattern mining. Extensive experimental evaluation demonstrates that our algorithms provide a secure, practical, and feasible algorithm for mining frequent patterns on a public cloud.