A New Privacy-Preserving Data Mining Method Using Non-negative Matrix Factorization and Singular Value Decomposition

The data analysis and mining is more and more powerful with the rapid growing data size. And publishing data for researchers is becoming more valuable. This process has an important problem: privacy protection. In recent decades, many methods for protecting privacy in data publishing have been studied. One important kind of them is based on matrix decompositions. These methods find non-critical information for analysis task using matrix decompositions and remove it from the data to protecting privacy. This paper improves this kind method and gives a new algorithm for protecting privacy based on non-negative matrix factorization and singular value decomposition. Our basic idea is that if using plurality kinds of decompositions, it can analyze data from different directions and will analyze data more comprehensive. So, it may find more non-critical information and improve the algorithm performance. The experiments confirmed this idea. This new method can get better result than the traditional ones in which only one kind decomposition is used. Our method gives more powerful guarantee for protecting privacy when maintaining data quality.