A Frequent Itemset Mining Method Oriented to Differential Privacy Protection

Abstract

The invention discloses a differential privacy protection-oriented frequent item set mining method, comprising the following steps: calculating the support degree of all item sets, selecting frequent item sets therefrom; counting the length of each transaction in the data set, and calculating the truncated length After L, truncate the data set; calculate the upper limit of the number of frequent itemsets m and the number of frequent items λ, and construct a set F composed of frequent items according to the value of λ; construct the largest frequent itemset MFI set B and candidate item set C; Use set B to add noise to the itemsets in set C; use the initial MFI set B to calculate the support of each candidate item set, and then calculate the sum E of the error with the true support; search for B and B in B, use B replaces B and updates the value of the error sum E; when the error sum no longer decreases, stop the iteration and output the result. The invention can well prevent personal privacy leakage caused by publishing frequent item sets, and at the same time, the operation of truncating data sets can effectively improve the usability of mining results.

Description

technical field [0001] The invention relates to a frequent item set mining method for differential privacy protection, which belongs to the technical field of information security. Background technique [0002] With the rapid development of cloud computing and big data, data mining technology has made great progress in some in-depth research and applications. Frequent itemset mining is one of the core problems of data mining. Its goal is to find frequent itemsets in the data set, and it is widely used in real life. Although it can provide valuable information, publishing frequent itemsets without processing may cause serious personal privacy leakage. How to improve the availability of frequently published itemsets while protecting personal privacy has become one of the urgent problems to be solved in the field of data mining. With the proposal and development of privacy protection technology, differential privacy protection method has become a popular privacy protection te...

AI Technical Summary

Problems solved by technology

Differential privacy is realized through the noise mechanism, that is, random noise is added to the output to protect data security. The greater the noise added, the more secure the data, however, the lower the availability of the data, and vice versa

[0003] At present, there are many frequent itemset mining algorithms that satisfy differential privacy (such as PrivBasis algorithm, PrivSuper algorithm, etc.), but these algorithms are only suitable for processing low-dimensional data sets, and they will be too sensitive when processing high-dimensional data sets. Add more noise to the problem, resulting in low usability of mining results

Images