Efficient multi-party privacy preserving data mining for vertically partitioned data
Surbhi Sharma, Deepak Shukla · 2016
The data in computational domain stored in digital format. This format of data, consumes less effort and storage. Thus a number of organization and institutes are preserving their information in this format. In this presented work the data and their privacy is the main area of study. In the proposed work an organization is considered where the decisions are made with the different department based data and their attributes. Additionally to make decisions the attributes of all the departments are required. But the departments are not able to disclose the privacy of data owner. Therefore to combine the data attributes and mining of the combined data need a privacy preserving technique for preventing the privacy issues in a centralize database. In this paper we demonstrate how the different departments of same organization combine their data without harming the privacy of the client for making effective decisions in efficient and accurate manner. Thus the method vertically data combination, cryptography and decision mining is demonstrated. To mine the decisions from the data a C4.5 decision tree is used. The implementation of the proposed privacy preserving data mining and decision making technique is performed using JAVA technology. Additionally the performance of the system is computed in terms of accuracy, error rate, memory consumption and time consumption. Finally to justify the outcomes of the proposed data mining technique the traditional J4.5 tree using WEKA tool is used with same data for comparative performance study. The experimental results show the effective performance and security in the given privacy preserving technique.