Investigation of privacy preserving data publication using attribute based clustering and encryption techniques

Abstract

Privacy preservation plays an important role in protecting sensitive data when sharing the information to public users in data mining. Privacy Preserving Data Publishing (PPDP) anonymizes the data through preserving the identity of individuals and sensitive information. Dimensionality reduction is a method of providing the high privacy rate for individual data within the database. Many research works have been done for preserving the data privacy on high dimensional data. However, nonymization technique reduces the privacy level and data utility due to the presence of various attacks while preserving the sensitive information for newlinedata distribution. Therefore, this research work focused on developing the privacy preservation through anonymization with reduced time complexity and higher anonymity level for efficient data publishing on high dimensional database. newlineAn accuracy constrained privacy preserving access control framework was planned for increasing the privacy level and access control based on the k-anonymous Partitioning with Imprecision Bounds (k-PIB). However, time complexity was minimized. Slicing technique was designed for preserving the attribute disclosure and also evaluates the sliced data based on l-diversity requirement. Though, the data utility was not sufficient due to random generation of connections between column values of bucket on high dimensional data. In order to improve the data utility, an anonymization technique was developed based on Nearest-Neighbor (NN) search and global data reorganization on sparse high-dimensional data. This in turn reduces the information loss with high privacy. newline newline