Clustering-anonymity method for data-publishing privacy preservation

摘要

Data-publishing generally need to be treated by anonymity to protect its privacy information from disclosure. Existing anonymity methods have little distincation between different types of Quasi-identifiers in investigating generalization. Aimed to privacy preservation for pulblishing data from table, A clustering-anonymity data publishing method is proposed by using the ideas of clustering algorithm. The method makes generalization into Quasi-identifiers according to its different type, It gives the reasonable definition of the distance between one tuple and the other or one equvialance class; Dueing to partitioning cluster one by one controlled by the value of k, it achieves partition with the approximate same size of every equvialance class, So it reduces the amount of calculation of distances, and saves the running time accordingly. Experimental results verify the effectiveness of the method.