Research On Privacy-preserving Association Rule Mining Based On Homomorphic Encryption

With the development of big data technology and cloud computing technology,data mining technologies play an important role to produce huge market values.Relying on the powerful computing capabilities of cloud servers,large-scale user data are collected by cloud service providers to provide association rule mining services for miners with limited computing resources.According to the mining results,the mall can formulate effective promotion strategies,and the website can reasonably recommend web pages to users.Although outsourcing association rule mining can provide consumers with convenience while improving corporate profits,there is also a risk of sensitive information leakage of cloud service users.Therefore,formulating a safe and effective privacy-preserving association rule mining scheme is very in lines with actual needs.It is required that it cannot only obtain the correct mining results but also protect the privacy of data owners and data miners.The main work done in this article is as follows: 1.We propose a variant-Paillier homomorphic encryption algorithm and a privacypreserving association rule mining scheme for large shopping malls in a multi-key environment.So far,there are not many homomorphic encryption schemes suitable for the construction of privacy-preserving data mining in a multi-key environment.In view of this phenomenon,we propose a variant-Paillier homomorphic encryption algorithm with a double decryption mechanism.each user in the scheme is allowed to have his own public and secret key pair by this encryption.Compared with the BCP homomorphic encryption algorithm that also has a double decryption mechanism,our algorithm can reduce the size of the ciphertext and improve transmission efficiency.In addition,we construct a privacypreserving association rule mining scheme in a multi-key environment based on the proposed homomorphic encryption.Considering the large variety of goods in large shopping malls,using binary vectors to represent transaction records can lead to a large amount of invalid storage.This scheme uses vectors of item numbers to represent transaction records,which can reduce storage and reduce the amount of redundancy in calculations.Based on the representation method,a set of secure computing protocol are designed.Through analysis,our proposed scheme can not only ensure the correct mining results but also ensure the privacy of the mining process.2.We propose a privacy-preserving association rule mining scheme under a single cloud server structure.In the existing privacy-preserving association rule mining schemes,the public key cryptography-based schemes achieve high privacy security and accuracy of results at the same time,but they are mostly based on the structure of two non-collusion cloud servers.In the case of only one cloud server,these existing schemes cannot safely implement association rule mining.Therefore,we propose a privacy-preserving association rule mining scheme based on homomorphic encryption in a single cloud server structure.This scheme uses the Paillier homomorphic encryption algorithm,the garbled circuit and hash algorithm to design a secure multiplication protocol a single cloud server structure.Based on this protocol,it can safely calculate the ciphertext of the inner product of the query vector and the record vector.According to the designed subprotocol,a security association rule mining protocol is given.Also,we analyzed the security of the protocol against possible threat models and proved the security of the scheme.At the end of the scheme,we performed an experimental simulation of the scheme performance.