Compatible Data Anonymization for Multi-Source Privacy and Utility
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for anonymizing data sets from multiple sources struggle to maintain statistical representativeness and privacy while ensuring compatibility, especially when data is sensitive and cannot be shared directly.
Innovation Solution
A method using machine learning equipment to define data types, identify quasi-identifiers, and apply data transformations to anonymize data sets, optimizing privacy metrics and minimizing information loss through trained models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If data sets from different sources are combined for analysis, then the quantity and variety of data increases, but the risk of re-identification and privacy breaches increases
Solution Approach 1:
The patent applies parameter changes by transforming data through k-anonymity generalization (modifying data parameters to hide identifying characteristics) and adding controlled noise to statistical queries. This allows the system to maintain data utility for analysis while changing parameters sufficiently to prevent re-identification, thus resolving the contradiction between data quantity and privacy protection.
Solution Approach 2:
The patent introduces an intermediary mechanism - a privacy-preserving data processing system that acts as a mediator between raw multi-source data and analysis queries. This intermediary applies anonymization algorithms and controlled noise addition, allowing data to be combined for analysis while preventing direct access to identifying information, thus enabling both data quantity increase and privacy protection.
2Reliability
If data is anonymized to protect privacy, then privacy protection improves, but data utility and analytical value deteriorate
Solution Approach 1:
The patent applies partial action by implementing k-anonymity with k≥2, which provides sufficient privacy protection without excessive generalization that would destroy data utility. The system also applies controlled noise addition - enough to prevent inference attacks but not so much as to render statistical analysis meaningless, thus balancing privacy protection with data utility preservation.
Solution Approach 2:
The patent uses parameter changes through selective generalization - modifying only the identifying parameters (quasi-identifiers) while preserving the analytical parameters. This allows the data to maintain its analytical value for research purposes while changing the identifying parameters sufficiently to protect privacy, thus resolving the contradiction between privacy protection and data utility.
3Reliability
If multiple data processing operations are applied to ensure privacy, then privacy protection improves, but processing complexity and computational cost increase
Solution Approach 1:
The patent applies preliminary action by pre-computing and storing anonymized versions of data sets before actual analysis queries are executed. The k-anonymity generalization and noise parameters are predetermined and applied in advance, reducing the computational complexity during actual query processing. This allows multiple privacy-preserving operations to be performed once rather than repeatedly for each query, thus reducing processing complexity while maintaining strong privacy protection.
Data Source
Figure 1~2
Figure 3
AI summary
A method, computer program and machine learning equipment for creating compatible anonymized data sets by automatically performing with machine learning equipment that operates a machine learning model: defining data types of individual variables of a first data set; identifying quasi-identifiers for the first data set; defining reidentification sensitivity of all or any targeted subset of the individual variables and quasi-identifiers; defining missing data handling rules for the individual variables; defining allowed data transformations including generalization and use of synthesized data; optimizing quasi-identifier selection, use of synthesized data and a choice of data transformations to minimize information loss and maximize privacy metrics based on at least all of: the first data set; the allowed data transformations; and the missing data handling rules; training the machine learning model using: the first data set according to the defined data types; the optimized quasi-identifier selection; the optimized use of synthesized data; and the choice of data transformations; and anonymizing the first data set using the training of the machine learning model.