Differential privacy protecting method of datastream key mode mining

Abstract

The invention discloses a differential privacy protecting method of datastream key mode mining. A window corresponding to each timestamp is preprocessed and mined, firstly first noise adding is conducted, so that counting (a frequent mode) satisfies differential privacy, and then through a branch number distribution condition, all nodes are sifted to select a node candidate set C1 which satisfiesthe condition that c(primary set)>Sigmac(subsidiary set), and nodes in the C1 are subjected to secondary noise adding, so that the key mode satisfies the differential privacy. By conducting noise adding twice, on the premise that the safety is satisfied, the effectiveness can be improved. Besides, for window sliding, to improve reasonable distribution and privacy budget of the mining time, two differential nodes are utilized to calculate and decide noise adding situations, counting is considered, and the branch number condition is also considered, so that the mined key mode lowers the computing complexity as much as possible on the condition that the safety is guaranteed and is for study personnel to conduct analysis work.

Description

technical field [0001] The invention relates to the technical field of privacy protection, in particular to a differential privacy protection method for mining key patterns of data streams. Background technique [0002] With the advent of the information age, the information technology industry has entered a stage of rapid development. The Internet has penetrated into every field of human activities. The development of the Internet has produced a large amount of data in various fields such as politics, economy, culture, medical care and education, and mining these data can obtain a lot of useful information. Frequent Itemset Mining (FIM) is a core task in data mining and has been extensively studied. Key patterns as the best subset of frequent patterns have played a great role in many fields, such as decision support, Web usage mining, bioinformatics, and so on. For a stream of transactions, where each transaction contains a set of items, FIM tries to find sets of items t...

AI Technical Summary

Problems solved by technology

[0008] What the present invention aims to solve is that the existing differential privacy protection method has the problem of privacy leakage when facing the mining of key patterns under the data flow, and provides a differential privacy protection method for mining key patterns of the data flow

Images