Minimization of Data Distortion on a Privacy Protection Technique against Attacks Using Background Knowledge
Shunsuke Muramoto · 2008
In this paper, we consider a privacy protection technique to convert an input data table in a database into one maintaining l-diversity by generalizing data. In the previous work, we proposed an algorithm which pre- vents data guess by the data combination by letting data table maintain k-anonymity, and outputs result data tables with small data distortion degree (a degree of dierence between original data and result data). Recently, however, there exists a new kind of attacks that we cannot prevent even with k-anonymity. Therefore, Machanavajjhala et al. introduced a new property as the l-diversity to prevent such new kinds of attacks, and we improve our algorithm for privacy protection against the new attacks so as to output result tables by minimal distortion.