A Joint-Entropy-Based Anonymity Metrics Model with Multi-Property

An anonymity metrics model with three properties based on joint entropy is given in this paper. The three properties are identifiable, linkable and traceable respectively. According to randomness and fuzziness of anonymity, a fuzzy pattern recognition model is presented based on entropy and least generalized weighted distance, which makes the membership vector of anonymity grades have a desirable dispersive property. A method is proposed for computing the balance parameter between entropy and the generalized weighted distance. It is illustrated that the presented method has advantages over Shannon entropy like models in performance. The method can determine uniquely the balance parameter defined in this paper. Therefore, joint entropy is applied to evaluate the anonymity grades.