Research On Anonymous Method Based On Proximity Resistance Of Sensitive Information

Data anonymization is often used to solve the problem of privacy leakage in data publishing due to its unique advantages in security and effectiveness.aiming at the problem of privacy leakage caused by similarity attacks,this paper proposes(r,k)-anonymity model,sets proximity resistance threshold based on proximity association of sensitive attributes,and designs an anonymous algorithm GDPPR(Generalized data for privacy proximity resistance)that satisfies(r,k)-anonymity model.the fuzzy clustering technique based on(r,k)-anonymous model improvement is used to complete the cluster division,and the membership matrix is obtained,so the frequency of sensitive attributes with adjacent associations in each equivalence class is not higher than the threshold After anonymization is ensured,the data can meet the user's privacy protection needs and it is valuable to use.this article is mainly based on the following two key points:1.Research the anonymous model of sensitive attribute proximity resistance and design an anonymous algorithm that satisfies(r,k)-anonymous model GDPPR.aiming at the problem that similarity attacks will cause privacy leakage,this paper proposes(r,k)-anonymous model.it is required that the data table after anonymization meets the k-anonymity model,and the frequency of sensitive attributes with adjacent associations in the same equivalence class does not exceed the threshold .effectively reduces the probability of attackers obtaining similar sensitive information through similarity attacks and verifies(r,k)-the anonymous effect of the anonymous model through experiments.2.Research the availability of data in the GDPPR algorithm.the GDPPR algorithm completes the division of clusters by fuzzy clustering technology and then effectively improves data availability during the process of generalization to form equivalent classes.in view of the problem of unnecessary information loss caused by generalization of attributes,this paper improves fuzzy clustering technology based on(r,k)-anonymous model,and completes the clustering of data sets according to the improved fuzzy clustering.through experimental comparison,it is proved that the data set after clustering can effectively reduce information loss and improve data availability in the process of forming equivalence classes.In this paper,two standard data sets of the Adult data set and the Census-Income data set in the UCI machine learning library were selected to conduct the experimental comparison tests.the results prove that the GDPPR algorithm can well satisfy the(r,k)-anonymous model and can effectively resist similarity attacks.and it has better anonymity and data availability under the similar time cost as Mondrian algorithm.