A New Hybrid Approach For Privacy Preserving Data Mining Using Matrix Decomposition Technique

With the increase of data size, data analysis is becoming very important for finding hidden knowledge. This data may contain sensitive information of individuals that could be revealed while accessing the data from sources. Most individuals or organizations do not want to reveal their personal information because of privacy leakage. In recent years, many models and methods have been designed for preserving privacy such as using matrix decomposition techniques. In this paper, Sparsfied Singular Value Decomposition (SSVD) and Non-negative Matrix Factorization (NMF) are used and a hybrid approach based on those methods has been proposed. The main reason is that the decomposition of matrix more than once removes more sensitive information. By removing the sensitive data, we can improve the privacy of the dataset and keep the data utility considerable. The approach provided here ensures better privacy while keeping good data utility.