Uncertain data frequent item set publishing method based on differential privacy

Abstract

The invention discloses an uncertain data frequent item set publishing method based on differential privacy, which comprises the following steps: (1) giving a data set D, carrying out frequent item set mining on the D to obtain the support degree [lambda] of a frequent 1-item set ILIST and a kth frequent item set, and adding noise to obtain the support degree [lambda] of the frequent 1-item set ILIST and kth frequent item set; (3) extracting a length estimation value of a maximum frequent item set containing each item x from the D', and taking the length estimation value as hierarchical information; and (4) setting an initial node as a null set in the D', and screening by using hierarchical information constraints and subset frequent constraints to obtain a candidate item list NodeList ofthe current node; and (5) screening a top-k frequent item set and a support degree thereof in the FISet, and publishing the topk frequent item set and the support degree thereof. The whole process ofthe method meets differential privacy, and the privacy security of data can be improved.

Description

technical field [0001] The invention relates to a method for publishing frequent itemsets of uncertain data based on differential privacy, which belongs to the publishing algorithm of frequent itemsets in the field of uncertain data. Background technique [0002] At present, the research on the release of frequent itemsets of privacy-preserving uncertain data is still in its infancy, and the existing research is mainly based on differential privacy. Differential privacy has become a new popular privacy protection model, which does not care about the background knowledge possessed by the attacker. All data. Differential privacy achieves privacy protection by adding noise to query or analysis results. Existing methods for publishing frequent itemsets of uncertain data with privacy protection based on differential privacy have the following shortcomings: (1) The privacy protection mechanism is independent of the mining process, which causes the accuracy of the published top-k...

AI Technical Summary

Problems solved by technology

The existing methods of publishing frequent itemsets of privacy-preserving uncertain data based on differential privacy have the following deficiencies: (1) The privacy protection mechanism is independent of the mining process, causing the accuracy of the published top-k frequent itemsets to be affected by the k value (2) The algorithm uses Exponential mechanism, when applied to large-scale frequent itemsets, the privacy budget is large, and it is difficult to take into account data availability and privacy security (3) Separating the association between frequent itemsets and their support for privacy protection processing, resulting in unsatisfactory output results Top-k descending order constraints and correctness constraints

In summary, the existing methods have the defect that the privacy protection mechanism is independent of the mining process, and it is difficult to balance privacy protection security and data availability.

Images