Visible to the public A service-oriented adaptive anonymity algorithm

TitleA service-oriented adaptive anonymity algorithm
Publication TypeConference Paper
Year of Publication2020
AuthorsZhang, M., Wei, T., Li, Z., Zhou, Z.
Conference Name2020 39th Chinese Control Conference (CCC)
Date PublishedJuly 2020
ISBN Number978-9-8815-6390-3
KeywordsAdaptation models, anonymity, anonymity group partition process, anonymized data, anonymous results, attribute value distribution, attribute values distribution, composability, contribution value differences, data handling, data privacy, Dispersion, distribution characteristics, Human Behavior, Internet of Things, k-anonymity, K-anonymity algorithms, Metrics, Partitioning algorithms, privacy, privacy preservation, privacy-preserving data publishing, pubcrawl, Publishing, Quasi-identifier Attributes, quasiidentifier attributes, released data, resilience, Resiliency, sensitive attributes, service-oriented adaptive anonymity algorithm, service-oriented architecture, splitting value

Recently, a large amount of research studies aiming at the privacy-preserving data publishing have been conducted. We find that most K-anonymity algorithms fail to consider the characteristics of attribute values distribution in data and the contribution value differences in quasi-identifier attributes when service-oriented. In this paper, the importance of distribution characteristics of attribute values and the differences in contribution value of quasi-identifier attributes to anonymous results are illustrated. In order to maximize the utility of released data, a service-oriented adaptive anonymity algorithm is proposed. We establish a model of reaction dispersion degree to quantify the characteristics of attribute value distribution and introduce the concept of utility weight related to the contribution value of quasi-identifier attributes. The priority coefficient and the characterization coefficient of partition quality are defined to optimize selection strategies of dimension and splitting value in anonymity group partition process adaptively, which can reduce unnecessary information loss so as to further improve the utility of anonymized data. The rationality and validity of the algorithm are verified by theoretical analysis and multiple experiments.

Citation Keyzhang_service-oriented_2020