A new method for privacy-preserving data mining based on weighted singular value decomposition

作者：Li Guang^*; Wang Yadong

来源：Journal of Convergence Information Technology, 2011, 6(3): 28-34.

DOI：4156/jcit.vol6. issue3.4

• Data sample
• Data utilities
• Privacy preservation
• Privacy protection
• Privacy-preserving data mining
• Weighted SVD

Privacy protection is indispensable in data mining, and many privacy-preserving data mining (PPDM) methods have been proposed. One such method is based on singular value decomposition (SVD), which uses SVD to analyze and to perturb the original data and open the modified data to users instead of the original data. In the SVD-based method, all data samples are treated equally. However, because different samples do not hold the same importance for data mining, it is better to pay more attention to the important samples. This paper improves the SVD-based method using this idea and presents a new PPDM algorithm based on a weighted SVD. In this new method, each sample has a weight, and different samples will be treated with different weights. Our experiments show that while maintaining the data utility, our new weighted SVD-based method is significantly improved over the original SVD-based method in privacy protection.

• 出版日期2011