PRIVACY PRESERVING NAIVE BAYES CLASSIFIER FOR HORIZONTALLY PARTITIONED DATA

Abstract

The data mining process can reveal sensitive information about individuals or organizations thereby violating their privacy. The main purpose of the field of privacy preserving data mining is to develop various techniques to find valuable knowledge or information while still keeping sensitive data and private information for the owners. Up to now, there have been many solutions proposed, however these solutions either have low efficiency or do not ensure privacy. This article builds a privacy preserving Naive Bayes classifier solution in a multi-member classifier for horizontally partitioned data scenario based on the application of the secure sum protocol. The proposed protocol is assessed as good privacy, accuracy and efficiency in comparison to contemporary solutions. To confirm the effectiveness of the proposed solution, in the experimental part, the author used the python programming language to visualize the results. The author specifically, build a privacy preserving Naive Bayes classifier solution for the spam message detection model. Experimental results show that the proposed solution has good applicability in practice.