(a, d)-Diversity: Privacy Protection Based on l-Diversity

Abstract

In recent years, privacy protection has been emphasized while publishing data with sensitive information. Existing proposals for privacy protection can well avoid identity disclosure; however they do not provide sufficient protection for privacy under background knowledge attack. This paper analyzes the cause of attribute disclosure and proposes a novel idea for privacy protection based on l-Diversity. It takes the semantic meaning of the sensitive attributes into consideration and gives a stronger definition of privacy protection. First, the sensitive attribute values are divided into groups, and then the records are grouped according to the sensitive attribute. Finally, the table is anonymized. The experiment results shown in the paper demonstrate the feasibility of the proposal.