Privacy-preserving naive Bayes classification on distributed data via semi-trusted mixers

作者:

Highlights:

摘要

Distributed data mining applications, such as those dealing with health care, finance, counter-terrorism and homeland defense, use sensitive data from distributed databases held by different parties. This comes into direct conflict with an individual's need and right to privacy. It is thus of great importance to develop adequate security techniques for protecting privacy of individual values used for data mining.In this paper, we consider privacy-preserving naive Bayes classifier for horizontally partitioned distributed data and propose a two-party protocol and a multi-party protocol to achieve it. Our multi-party protocol is built on the semi-trusted mixer model, in which each data site sends messages to two semi-trusted mixers, respectively, which run our two-party protocol and then broadcast the classification result. This model facilitates both trust management and implementation. Security analysis has showed that our two-party protocol is a private protocol and our multi-party protocol is a private protocol as long as the two mixers do not conclude.

论文关键词:07.05.Kf,Privacy-preserving distributed data mining,Classification,Data security

论文评审过程:Received 15 December 2006, Revised 10 June 2008, Accepted 17 November 2008, Available online 6 December 2008.

论文官网地址:https://doi.org/10.1016/j.is.2008.11.001