Privacy protection on multiple sensitive attributes

Abstract

In recent years, a privacy model called k-anonymity has gained popularity in the microdata releasing. As the microdata may contain multiple sensitive attributes about an individual, the protection of multiple sensitive attributes has become an important problem. Different from the existing models of single sensitive attribute, extra associations among multiple sensitive attributes should be invested. Two kinds of disclosure scenarios may happen because of logical associations. The Q&S Diversity is checked to prevent the foregoing disclosure risks, with an α Requirement definition used to ensure the diversity requirement. At last, a two-step greedy generalization algorithm is used to carry out the multiple sensitive attributes processing which deal with quasi-identifiers and sensitive attributes respectively. We reduce the overall distortion by the measure of Masking SA. © Springer-Verlag Berlin Heidelberg 2007.