Research On Methods Of Privacy Preserving Forassociation Rule Mining In Distributed Environment

Along with the rapid development of network, communications and computer technology, data mining takes on the features of large datasets and distributed mining. How to preserve data privacy and prevent sensitive information from disclosure has become a great challenge. Privacy preserving data mining (PPDM) has become an increasingly important issue in data mining data mining(DM) field. Association rule mining is one of the most widely used data mining methods. In this thesis, the methods of privacy preserving for association rule mining in distributed environment are studied, the purpose of the study is to maximize mining the potential knowledge in the database and protect data privacy at the same time.This thesis introduces several privacy preserving technologies which are commonly used in data mining from two aspects: the protection of sensitive data and the protection of sensitive knowledge. On the basis of the overview of data mining technology, the thesis introduces theories of distributed association rule mining, analyzes several popular existing distributed association rule mining algorithms and their advantages and disadvantages. Then, the thesis has done a lot of further research work for preserving sensitive knowledge in distributed association rule mining, which are as follows:(1)For the horizontally partitioned datasets, a data cleaning algorithm to hide sensitive rules in local site is proposed. This algorithm changes the datasets little and realizes perfect hiding, while maximizing the accuracy of the results of the global mining and the effects of hiding sensitive rules. And a scheme of combining the RSA and HES is adopted to encrypt the information of frequent itemsets that transport between sites, the security of data encryption and the efficiency of encryption algorithm are considered in this scheme, and it reaches the balance between efficiency and security.(2)For the vertically partitioned datasets, the key for privacy preserving in distributed association rule mining is to compute global frequent itemsets safely. A new protocol to securely compute the support of the itemsets is put forward; it can compute the support of the itemsets accurately and protects the private information of each site.(3) In order to verify the algorithms designed by this thesis, experiment is done, and the experimental results show that those algorithms can achieve improvements in terms of privacy, accuracy, and efficiency.The thesis has done beneficial research work on privacy preserving for association rule mining in distributed environment.