Anonymous Privacy Protection Algorithm Based on Sensitive Attribute Classification

Abstract

At present, most personalized privacy protection algorithms can be divided into two methods for protecting sensitive attributes. One is to set different thresholds for different sensitive attributes; the other is to generalize sensitive attributes, and replace the original sensitive attribute values with lowprecision generalized values. The anonymized data of the two methods has the risk of sensitive information leakage or large information loss, as well as the problem of data availability. To this end, a personalized (α, p, k) anonymous privacy protection algorithm is proposed. According to the sensitive level of the sensitive attribute, different anonymous methods are adopted for the sensitive values of each level in the equivalence class, so as to realize personalized privacy protection of the sensitive attribute. Experiments show that this algorithm has an approximate time cost and lower information loss than other personalized privacy protection algorithms.