Private Data Sampling via Multi-Party Computation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data sampling techniques, such as rejection sampling, fail to adequately protect sensitive information and privacy in remote dataset sampling, as they require disclosure of the sampling criterion to data providers, potentially leading to data leakage or increased costs.
Innovation Solution
A data sampling device and sample providing device utilize multi-party computation to sample records from a remote dataset while keeping the sampling criterion private, using a candidate sample and multi-party computation to determine a subset of indices and records that satisfy the criterion, reducing the need for information disclosure and minimizing data leakage risks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If rejection sampling is performed by data providers on behalf of researchers, then sampling efficiency is improved, but the researcher must disclose the criterion to data providers, leading to potential data leakage and loss of research direction privacy
Solution Approach 1:
The patent introduces a cryptographic intermediary layer (secure multi-party computation protocol) between the researcher and data providers. This intermediary enables the researcher to perform rejection sampling without disclosing the criterion to data providers, while still allowing them to evaluate records. The criterion remains private to the researcher, eliminating information leakage while maintaining sampling efficiency.
2Loss of information
If rejection sampling is performed by the researcher himself, then criterion privacy is maintained, but the researcher needs access to additional information beyond the final dataset, increasing data exposure risks
Solution Approach 1:
The patent uses secure multi-party computation as a cryptographic intermediary that allows the researcher to evaluate the criterion on remote records without actually accessing or exposing the records themselves. The data providers can evaluate records against the criterion while the criterion remains private to the researcher, and neither party exposes sensitive information to the other, thus reducing data leakage risks.
3Object-affected harmful factors
If data providers keep additional information private to minimize data leakage risks, then data security is improved, but the researcher cannot verify the criterion evaluation, reducing sampling reliability
Solution Approach 1:
The patent implements a feedback mechanism through secure multi-party computation where the data provider evaluates records against the criterion and returns only the evaluation results (not the records or criterion). The researcher receives feedback in the form of acceptance/rejection decisions without learning the criterion or accessing the records, maintaining both security and reliability through cryptographic verification.
Data Source
AI summary
Some embodiments are directed to a data sampling device for obtaining a sample of records from a remote dataset satisfying a private criterion using multi-party computation. One or more sample providing devices store respective subdatasets of the remote dataset. The data sampling device determine a candidate size for a sample providing device; requests the sample providing device to determine a candidate sample of the candidate size from the subdataset of the sample providing device; perform a multi-party computation with the sample providing device to obtain a set of indices of records from the candidate sample satisfying the private criterion; sample a subset of the set of indices; and obtains from the sample providing device records of the candidate sample corresponding to the subset of the set of indices.