Privacy Preserving Data Mining Using Non-Negative Matrix Factorization and Singular Value Decomposition

With the increasing technologies, people share information time to time from different places. The rapid growing improved technologies and improved data mining algorithms make it easy for adversary to disclose sensitive information. So there is always a second thought while individuals share their personal information. That's why privacy protection is an important consideration at every stage of data mining process. Privacy Preserving Data Mining (PPDM) maintains data utility and protects privacy at the same time. In this paper, real world datasets are perturbed by using combined Non-Negative Matrix Factorization (NMF) and Singular Value Decomposition (SVD) method. The utility of perturbed datasets is analyzed with respect to query accuracy. The result of query accuracy implies that the combined method needs to be improved or more efficient methods need to be introduced.