Method and device for detecting relevance between user attributes

Abstract

The embodiment of the invention provides a method and device for detecting the relevance between user attributes, and is used for detecting the relevance between p user attributes for m sample users jointly provided by S data parties in service processing of the S data parties based on security calculation. The S data parties locally and respectively have part of user attribute data as private data. In order to maintain the security of private data of each data party, the method comprises the following steps of: splitting and calculating covariance matrixes corresponding to p user attributes by each data party, respectively determining each correlation coefficient corresponding to each two user attributes in the p user attributes according to the covariance matrixes, and detecting the relevance among the p user attributes based on each correlation coefficient and the covariance matrixes. Compared with the prior art that the VIF of each user attribute is directly calculated according tothe attribute value to determine the incidence relation of the user attributes, the method has the advantages that the calculation complexity can be greatly reduced, and the effectiveness of safety calculation is improved.

Description

technical field [0001] One or more embodiments of this specification relate to the field of computer technology, and in particular to a method and device for detecting correlations between user attributes in business processing based on multi-party secure computing. Background technique [0002] With the development trend of big data, data sharing has become a research value and practical issue. Data sharing usually refers to multiple data parties working together on data mining or machine learning while protecting their respective data privacy, in order to dig out more valuable content in the data. For example, three financial platforms (such as three banks) serve as three data parties, each of which has several users' credit data, which can be jointly trained to train a comprehensive credit model, which can be used as a reference for lending to users and other businesses. . [0003] However, multiple data parties in multi-party secure computing usually do not disclose th...

AI Technical Summary

Problems solved by technology

[0003] However, multiple data parties in multi-party secure computing usually do not disclose their private data to other data parties due to the security of private data

In the process of constructing training samples, there may be a correlation (such as linear correlation, also called collinearity) between the user attributes of the sample users provided by each data source, that is, one user attribute can be well passed by at least one other User attribute representation, resulting in feature redundancy of training samples

Images