Contextual Entropy Management for Re-Identification-Resistant Data Privacy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing de-identification methods are vulnerable to re-identification attacks and severely limit the utility of de-identified data, failing to provide secure and context-based privacy protection.
Innovation Solution
A contextual privacy system introduces entropy to data values using a probabilistic and dependent offset generation method, applying perturbation values based on privacy context levels to prevent re-identification while preserving data utility.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional de-identification methods are used to modify identifiable portions of data, then privacy protection is improved, but the data becomes vulnerable to re-identification attacks and utility is severely limited
Solution Approach 1:
The patent changes the parameter of perturbation from fixed to variable, where the magnitude of perturbation is dynamically adjusted based on the sensitivity of the data and the privacy requirements. This allows the system to apply stronger perturbation to highly sensitive data while preserving more utility for less sensitive data, resolving the contradiction between privacy protection and data utility
Solution Approach 2:
The patent introduces dynamic adaptability by making the de-identification process context-dependent. The system dynamically adjusts perturbation strategies based on data characteristics, query types, and privacy contexts, enabling the system to optimize both privacy protection and data utility for different scenarios rather than applying a static one-size-fits-all approach
2Reliability
If random value modifications are applied to de-identify data, then privacy is improved, but the system becomes vulnerable to iterative re-identification attacks
Solution Approach 1:
The patent implements feedback mechanisms where the system monitors and tracks perturbation patterns and queries. This feedback loop enables the system to detect potential re-identification attacks and dynamically adjust perturbation strategies to counteract them, making iterative attacks ineffective while maintaining privacy protection
Solution Approach 2:
The patent applies preliminary perturbation to data before it is accessed or queried. By pre-applying context-aware perturbation based on data sensitivity and anticipated privacy requirements, the system prevents attackers from gaining useful information through iterative queries, as the data is already protected with appropriate perturbation levels
3Reliability
If strong perturbation is applied to prevent re-identification, then privacy is improved, but data utility and accuracy are severely degraded
Solution Approach 1:
The patent applies local quality by differentiating perturbation strength based on data characteristics and context. Different data fields, records, or queries receive different perturbation levels tailored to their specific privacy risks and utility requirements, ensuring that privacy is protected where needed while preserving accuracy where possible
Solution Approach 2:
The patent dynamically changes perturbation parameters based on data sensitivity, query context, and privacy requirements. This allows the system to optimize the balance between privacy protection and data accuracy for each specific case, applying stronger perturbation only when necessary and maintaining higher accuracy for less sensitive or more utility-critical data
Data Source
AI summary
A method for context-based data privacy can include receiving an electronic file, determining that the electronic file includes a plurality of data variables that contain privacy-sensitive information related to at least one of a plurality of entities, causing presentation of a user interface specifying a plurality of fields of the electronic file on a computing device of a user, receiving a user selection of a subset of the plurality of fields each including one of the plurality of data variables, analyzing the electronic file using one or more rules to generate context data indicating a privacy privilege level for each field of the subset of fields, determining a de-identification process for each data variable in a respective field of the subset of fields based on one or more rules of the contextual privacy system and the privacy privilege level for each field of the subset of fields, and causing one or more data variables from the subset of fields to be modified using one or more respective de-identification processes.