Anonymization Module for Statistical Database Queries
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing anonymization methods for database queries are inefficient in computation and memory usage, and fail to effectively protect individual user privacy against condition/inverse condition pairs, often sacrificing data utility or being complex to configure.
Innovation Solution
A system and method using an anonymization module that identifies attack components in queries, adds noise elements to mask their effects, and ensures that statistical answers are perturbed in a way that makes it highly unlikely for information about individual users to be inferred, using fixed-random numbers and time epochs to generate noise, and ignoring database changes that affect the answer.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If noise is added to numerical answers to protect individual privacy, then individual privacy is improved, but data utility deteriorates due to reduced accuracy
Solution Approach 1:
The patent changes the parameter of noise addition from fixed to dynamic, adjusting the noise level based on query characteristics and data sensitivity. This allows the system to maintain stronger privacy protection when needed while preserving data utility when possible, resolving the contradiction between privacy and accuracy
Solution Approach 2:
The system dynamically adjusts the anonymization strategy based on query patterns, user roles, and sensitivity analysis. Rather than applying fixed noise to all queries, the system adapts the level of perturbation in real-time, allowing optimal balance between privacy protection and data utility for different scenarios
2Reliability
If complex anonymization methods are used to strengthen privacy protection, then privacy strength is improved, but system complexity increases
Solution Approach 1:
The patent segments the anonymization process into distinct modules: sensitivity analysis, query pattern detection, noise generation, and result perturbation. Each module handles a specific aspect of the anonymization task, making the overall complex system more manageable and easier to configure while maintaining strong privacy protection
Solution Approach 2:
The system introduces an intermediary anonymization layer between the database and the analyst. This intermediary automatically handles the complex transformations and noise additions, shielding the user from configuration complexity while enforcing strong privacy protections through the intermediary's built-in logic
3Reliability
If the number of queries is limited to prevent privacy loss, then privacy protection is improved, but productivity decreases
Solution Approach 1:
The patent implements periodic re-evaluation of query patterns and sensitivity levels, allowing the system to adjust anonymization parameters over time. This periodic action enables more queries to be processed within privacy budgets while maintaining protection, as the system learns from accumulated query data and optimizes the balance between productivity and privacy
4Reliability
If data values are modified to achieve K-anonymity, then individual privacy is improved, but data utility deteriorates due to loss of accuracy
Solution Approach 1:
The patent changes the approach from modifying data values (K-anonymity) to modifying query results (differential privacy). Instead of altering the underlying data which causes information loss, the system adds controlled noise to query answers, preserving the original data utility while achieving privacy protection at the output stage
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An anonymizing method for a database system is provided. The method includes receiving a query, identifying one or more attack components in the query, computing one or more noise elements from the one or more attack components, and computing fixed noise or fixed thresholds from the one or more noise elements.