Butterfly: Privacy Preserving Publishing on Multiple Quasi-Identifiers

Recently, privacy preserving data publishing has attracted significant interest in research. Most of the existing studies focus on only the situations where the data in question is published using one quasi-identifier. However, in a few important applications, a practical demand is to publish a data set on multiple quasi-identifiers for multiple users simultaneously, which poses several challenges. How can we generate one anonymized version of the data so that the privacy preservation requirement like k-anonymity is satisfied for all users? Moreover, how can we reduce the information loss as much as possible while the privacy preservation requirements are met? In this paper, we identify and tackle the novel problem of privacy preserving publishing on multiple quasi-identifiers. A naive solution of respectively publishing multiple versions for different quasi-identifiers unfortunately suffers from the possibility that those releases can be joined to intrude the privacy. Interestingly, we show that it is possible to generate only one anonymized table to satisfy the k-anonymity on all quasi-identifiers for all users without significant information loss. We systematically develop an effective method for privacy preserving publishing for multiple users, and report an empirical study using real data to verify the feasibility and the effectiveness of our method.