T-closeness privacy protection method satisfying epsilon-differential privacy

Abstract

The invention discloses a t-closeness privacy protection method satisfying epsilon-differential privacy. The method comprises the following steps: preprocessing original data (hospital patient data including names, ages, contact ways, native places and health conditions); establishing a quasi-identifier and sensitive attribute associated QIS data table, dividing the data table into a set of buckets according to a hierarchical tree of sensitive attributes in the data table, adding a dynamic programming algorithm to partition the data table, and finally generalizing according to a partition result to generate an anonymous data table. Aims at the defects that a traditional t-closeness privacy protection model cannot resist attackers with certain background knowledge, excessive dependence on sensitive attribute distribution and differential privacy may lead to loss of value significance of published data, a new distance function is introduced to optimize t-closeness, and a threshold t is optimized and adjusted, so that the combination of t-closeness and a differential privacy standard is realized, the data release effectiveness can be ensured on the premise of reducing data loss as much as possible, and the relationship between data is protected.

Description

technical field [0001] The data protection technology of the present invention specifically relates to a t-closeness privacy protection method satisfying ε-differential privacy. Background technique [0002] With the continuous development of Internet technology, human beings have gradually entered the era of big data, and the connection between various fields and network technology has become increasingly close, but at the same time it has also increased the risk of data privacy leakage. Since the information age, research on privacy protection has been on the rise. It started, and with the continuous growth of data, people pay more and more attention to privacy, and more and more privacy protection technologies have been proposed one after another. Privacy protection is to limit the risk of information leakage to an acceptable range. It is generally achieved by anonymizing the data table before publishing the data. Commonly used privacy protection methods include k-anonymi...

AI Technical Summary

Problems solved by technology

Therefore, the privacy leakage risk caused by adding a record to the data set is controlled within a very small and acceptable range, and the attacker cannot obtain accurate individual information by observing the calculation results, which can effectively deal with background attacks. However, differential privacy is a strict definition, which assumes that the attacker has as much background knowledge as possible, so in order to meet the requirements of differential privacy protection, the published results often introduce enough noise to make the data meaningless

Images