A local differential privacy data sharing method and system with multiple privacy budgets

Abstract

The invention relates to a multi-privacy budget local differential privacy data sharing method and system, and provides a framework for a user to determine a privacy budget factor in a local differential privacy application scene. Based on an existing local differential privacy protocol, a step of enabling the user to confirm own privacy factor is added; A data collector obtains buckets accordingto distribution of privacy factors submitted by all users, the users form groups according to bucket dividing results, disturbance data finally submitted by the users of each group can use the same privacy factor, the privacy factor can be smaller than or equal to the privacy factor originally submitted by the users, and therefore privacy requirements of the users are met. A data collector calculates a relatively accurate statistical result through methods such as maximum likelihood and the like, and the accuracy of the statistical result can be proved by mathematics. The method can be directly applied to existing local differential privacy algorithms such as RAPPOR and SH protocols, and is easy to implement.

Description

technical field [0001] The invention belongs to the field of computer science and information security technology. In the local differential privacy application scenario, a framework for the user to independently confirm the privacy budget factor ε is proposed, which allows the user to independently define their own privacy protection level, and can effectively prevent some Untrustworthy data collectors maliciously use the high budget factor ε to extract the real privacy data of users. It can be proved that the present invention can be used for statistical activities carried out by standard local differential privacy protocols, and has high accuracy and strong practicability. Background technique [0002] After rigorous mathematical proof, differential privacy is currently one of the strongest privacy protection laws. It works by masking the user's data with carefully tuned noise. When many users submit their own sensitive data, the data center adds noise to the data to ac...

AI Technical Summary

Problems solved by technology

In today's local differential privacy applications (Rappor protocol, SH protocol), the setting of the ε value is basically determined by the data collector, which means that the security of the user's own data is still not completely determined by itself, resulting in local differential privacy in the application process. There is a vulnerability that data collectors maliciously select too large ε in

Images