Attribute based diversity model for privacy preservation

Privacy is a sensitive issue in our society nowadays in which individuals are trying to achieve by keeping their sensitive data secured. However, the demand of sharing sensitive data has been increased lately due to various tremendous benefits that can be achieved. Anatomy model based on k-anonymity and l-diversity models was presented for privacy preservation in data publishing. However, it cannot prevent the background knowledge attack that can occur based on the link between sensitive and non sensitive attribute values like gender or age. This paper evaluates the disclosure risk based on gender and age for Anatomy and proposes S-Cluster Approach to avoid such kind of disclosure. Experiments show that S-Cluster avoids background knowledge attack that is based on gender or age and produces better data quality comparing to Anatomy.