Skyline-based data generalization method

Abstract

The invention discloses a Skyline-based data generalization method. The method comprises the steps of processing a data table according to a data release privacy protection standard 10-anonymity to obtain a re-identified risk quantity R of a policy, recording the risk quantity R as a threshold T, and determining a policy space {S,(R,U)} according to a value domain of a quasi-identifier attribute and the threshold T, wherein an R value of the policy comprised in the policy space {S,(R,U)} is not greater than the threshold T; filtering the policy space {S,(R,U)} by adopting epsilon-approximate Skyline to obtain candidate policy spaces {G,(R,U)}; and performing Skyline calculation on the candidate policy space {G,(R,U)} to obtain a recommended policy space {F,(R,U)}, wherein the recommended policy space {F,(R,U)} is a private policy space recommended for the data table. According to the method, the accuracy of privacy protection policy recommendation is improved through an enumeration full policy space; the coverage range of an RU space is wide; multilevel demands of a user are met; the threshold T is set and the privacy protection policies not meeting the requirements are filtered, so that the policy space generation time is shortened; and the filtering is performed by adopting the epsilon-approximate Skyline, so that the scale of the candidate policy spaces is further reduced.

Description

technical field [0001] The invention belongs to the field of privacy-protected data publishing, and more specifically relates to a Skyline-based data generalization method. Background technique [0002] In the digital information age, it is becoming more and more important to exchange and publish data among various groups (such as governments, enterprises, individuals, etc.). For example, hospitals in California generally need to submit some medical data for certain recovered patients. These data contain some sensitive information, and publishing them directly will reveal personal privacy. Example of "connection attack" mentioned by L. Sweeney. By connecting the patient information table and the voter information table through attributes (Age, Sex, Zipcode), it can be determined that Ahmed has the flu. Patient privacy has been compromised. This method of data distribution is not secure. Therefore privacy-preserving data publishing is proposed. It requires maximizing da...

AI Technical Summary

Problems solved by technology

[0006] In view of the defects of the existing technology, the purpose of the present invention is to solve the problem that the existing privacy protection method has a small RU space coverage, cannot meet the multi-level needs of users, and has poor performance in both the risk amount R and the information loss amount U. Technical problems of low accuracy, small range of recommended strategies, and slow algorithm convergence

Images