Systematic clustering method for l-diversity model

Abstract

Nowadays privacy becomes a major concern and many research efforts have been dedicated to the development of privacy protecting technology. Anonymization techniques provide an e±cient approach to protect data privacy. We recently proposed a systematic clustering1 method based on k- anonymization technique that minimizes the information loss and at the same time assures data quality. In this paper, we extended our previous work on the systematic clustering method to l-diversity model that assumes that every group of indistinguishable records contains at least l distinct sensitive attributes values. The proposed technique adopts to group similar data together with l-diverse sensitive values and then anonymizes each group individually. The structure of systematic clustering problem for l-diversity model is defined, investigated through paradigm and is implemented in two steps, namely clustering step for k- anonymization and l-diverse step. Finally, two algorithms of the proposed problem in two steps are developed and shown that the time complexity is in O(n^2/k) in the first step, where n is the total number of records containing individuals concerning their privacy and k is the anonymity parameter for k-anonymization.