Random Number Generation for Multidimensional Anonymization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for anonymizing personal data in databases fail to prevent impossible state transitions in both numerical and category attributes during the anonymization process, as they do not effectively generate random numbers according to a multidimensional probability distribution where the probability in an impossible area is zero.
Innovation Solution
A random number generation apparatus and method that utilize a first and second random number generating part, along with a determining part, to generate random numbers according to a corrected distribution function that ensures a probability of zero in a predetermined area by using a basic distribution function and a hole in D-dimensional space, employing techniques such as the rejection method and inverse function method.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional anonymization methods using random numbers are applied to numerical attributes, then privacy protection is improved, but impossible state transitions occur (e.g., persons appearing in sea areas)
Solution Approach 1:
The patent modifies the probability distribution parameters by introducing a corrected distribution function f'(x) that incorporates constraints. The random number generation uses transformed parameters (α, β, γ) derived from the corrected distribution, ensuring generated values fall within valid ranges. This parameter transformation approach prevents impossible states while maintaining privacy protection through controlled randomization.
Solution Approach 2:
The patent introduces a corrected distribution function f'(x) as an intermediary between the basic distribution and the final random number generation. This intermediary function acts as a filter that blocks generation of impossible values while preserving the privacy-protecting randomization. The function incorporates constraint information (valid ranges for attributes) to guide random number generation away from impossible states.
2Reliability
If conventional anonymization methods are applied to category attributes, then privacy protection is improved, but impossible state transitions occur (e.g., high school students purchasing alcoholic drinks)
Solution Approach 1:
For category attributes, the patent transforms the probability distribution parameters to reflect valid category combinations. The corrected distribution function adjusts parameters based on constraints (e.g., age-category relationships), ensuring randomization only produces valid category transitions. This maintains privacy while preventing impossible category assignments.
Solution Approach 2:
The corrected distribution function serves as an intermediary that incorporates domain knowledge about valid category combinations. It mediates between uniform randomization (for privacy) and constraint satisfaction (for validity), allowing anonymization to proceed while blocking transitions to impossible category states through the constrained probability distribution.
3Ease of operation
If random numbers are generated without considering multidimensional constraints, then generation simplicity is maintained, but probability in impossible areas becomes non-zero
Solution Approach 1:
The patent segments the random number generation process into distinct steps: generating candidate values from a basic distribution, evaluating them against the corrected distribution function that incorporates constraints, and selecting final values. This segmentation allows the system to maintain simplicity in the basic generation step while achieving precision through the constraint-based correction step.
Solution Approach 2:
The patent performs preliminary action by pre-defining the corrected distribution function with all constraints embedded before random number generation begins. The valid ranges and category relationships are established in advance, allowing the generation process to simply sample from the pre-configured constrained distribution rather than checking constraints during generation.
Data Source
Figure 1
Figure 2
Figure 3(a)~3(b)
AI summary
A random number generation apparatus comprises: a first random number generating part 2 generating a random number u=(u1,...,uD)T ∈ [-∞,∞]D; a second random number generating part 3 generating a random number v ∈ [0,f'max]; and a determining part 4 determining whether f'(x1=u1,...,xD=uD)≥v or not, and, if f'(x1=u1,...,xD=uD)≥v, adopting u as a random number according to f'(x1,...,xD), wherein D is a predetermined positive integer, for i=1,...,D, [hi] is a predetermined possible range for a random variable xi, a hole [h] is [h]=([h1],...,[hD])T, H is a probability of a predetermined basic distribution function f(x1,...,xD) in the hole [h], α=1/(1-H), a corrected distribution function f'(x1,...,xD) is defined by Expressions (1) and (2), and fmax is a maximum value of f'(x1,...,xD).