A Batch Linear Query Method Based on Differential Privacy

Abstract

A batch linear query method based on differential privacy, including the following steps: Step 1: Query the original data set R to obtain a data query result set M; Step 2: Arrange the attribute frequency of R in descending order, and the screening frequency is not greater than the minimum support attribute and discard the attribute and the data corresponding to the attribute; perform data-independent processing on the attribute greater than the minimum support degree, and obtain an irrelevant data set D whose attribute frequency is greater than the minimum support degree; Step 3: use M in On the basis of establishing the initial load matrix, establish a data-independent load matrix W, and use parallel gradient descent matrix decomposition technology to decompose W in parallel to obtain the first matrix B of the complete decomposition result of W and the second matrix L of the decomposition result; Step 4: Based on Differential privacy performs adaptive noise addition, adds Laplacian noise to L and D, and restores the discarded attributes and data to obtain a noise-added query result dataset S; Step 5: Return S to the user.

Description

technical field [0001] The invention relates to the field of computer technology, in particular to a batch linear query method based on differential privacy. Background technique [0002] With the development of the Internet, human beings have entered the era of big data. When processing big data, batch linear query is the most commonly used operation. However, the query scale is huge, the query process is complicated, and the performance is low. In addition, in the process of using big data, a lot of sensitive information is easily leaked, and it is impossible to guarantee query accuracy (data availability) and privacy protection at the same time. [0003] Algorithms in the prior art cannot guarantee the algorithmic performance, query accuracy and privacy protection degree of batch linear query at the same time. In terms of algorithm performance, the existing algorithms have high complexity and are not suitable for large-scale batch linear queries; in terms of query accur...

AI Technical Summary

Problems solved by technology

When processing big data, batch linear query is the most commonly used operation, but its query scale is huge, the query process is complicated, and the performance is low

In addition, in the process of using big data, a lot of sensitive information is easily leaked, and it is impossible to guarantee query accuracy (data availability) and privacy protection at the same time.

[0003] Algorithms in the prior art cannot guarantee the algorithm performance, query accuracy and privacy protection degree of batch linear query at the same time

In terms of algorithm performance, the existing algorithm has high complexity and is not suitable for large-scale batch linear query; in terms of query accuracy, the existing algorithm adds noise to the query results to reduce the amount of noise required to optimize query accuracy

However, when the query sequence is arbitrarily given by the user, these mechanisms require a very large computational overhead in order to find the optimal noise distribution, which increases exponentially with the increase of the data dimension, and cannot be used for large data sets; in terms of privacy protection, existing The algorithm does not consider the amount of added noise combined with user permissions, and cannot guarantee that the amount of noise added to users with different permissions is appropriate. For users with high permissions, if too much noise is added, they will be greatly disturbed by noise and the query accuracy will decrease; For low-privilege users, adding too little noise will result in insufficient privacy protection

Images