K anonymity realization method and system

Abstract

The invention belongs to the technical field of data processing, and discloses a novel k anonymity realization method and system, and the method comprises the steps: adding noise to numerical attributes: dividing a whole data set into a plurality of equivalent anonymity classes, enabling the number of data in each anonymity equivalent class to be greater than k, and enabling the data in each equivalent class to be as similar as possible; Obtaining a range R of the equivalent anonymous class value attribute, and generating uniformly distributed random noise expected to be zero range R; Adding noise to each element of the anonymous class to realize k anonymity; K anonymity is realized on the non-numerical attribute through randomization; k anonymity is realized through generalization; And randomizing the generalized result. According to the method, the defect that several same quasi-identifiers appear in a traditional k anonymity method is overcome, and therefore better safety is achieved.

Description

technical field [0001] The invention belongs to the technical field of data processing, and in particular relates to a new method and system for realizing k-anonymity. Background technique [0002] At present, the existing technologies commonly used in the industry are as follows: [0003] There are two main types of methods to achieve traditional k-anonymity, clustering and generalization. [0004] Clustering is to classify the elements in the metadata set, at least the k closest elements, into one category, and use the quasi-identifier of the cluster center element to replace the identifiers of other elements in the class. This method is more suitable for numerical data. processing, but relatively poor for text data. [0005] Generalization is to expand the value of the specific quasi-identifier into a larger value range, so that it can no longer uniquely represent the unique tuple in the data set. For example, gender information can include male and female, which can be...

AI Technical Summary

Problems solved by technology

[0009] (1) In the traditional k-anonymous method, there will be several shortcomings of the same quasi-identifier, resulting in poor security;

[0010] (2) Cannot resist exhaustive attacks; no matter it is clustering or generalization, the quasi-identifiers of at least k records in the data set can be changed to the same value, making these records indistinguishable, thereby protecting the privacy of users

It is precisely because k records share the same quasi-identifier that user privacy is protected, but it is also precisely because k records share a common uniform identifier that there is a risk of privacy leakage. The current k-anonymity method, even if t is close to one requirements, and cannot resist brute force attacks

[0011] (3) The cost of realizing k-anonymity is relatively high; finding anonymous equivalent classes has always been a relatively time-consuming process in the k-anonymization process, and the calculation amount is large and time-consuming. The utility of the optimized data is greatly reduced

[0012] (4) The anonymization operation will lose part of the information of the original data set, making the availability of the anonymized data set relatively low

[0014] Finding an efficient privacy protection method is a prerequisite for the current informatization construction. Without an efficient and reliable privacy protection method, it is impossible for data owners to share the data they own. If data is not shared, data resources cannot exert their value. Therefore, it is of great theoretical and practical significance to find an efficient method that can protect privacy well.

[0015] A good privacy protection method needs to solve the following three problems, one is the problem of efficiency, the second is the problem of protection strength, and the third is the problem of ensuring data availability. The present invention proposes a high-efficiency, high-intensity privacy A method of protecting and, at the same time, guaranteeing the greatest possible availability of data

Images