Query Result Noise Hashing for Reverse-Engineering Privacy Protection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current noise addition techniques in data anonymization are susceptible to reverse engineering attacks, allowing malicious entities to infer sensitive information despite added noise, compromising individual privacy.
Innovation Solution
A method involving generating query results with noise based on a unique identifier, date, and record aspects, using random seed values and Gaussian distribution to adjust records, ensuring privacy through obfuscation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If unique identifiers are assigned to users or devices for tracking purposes, then tracking accuracy is improved, but user privacy is compromised
Solution Approach 1:
The patent introduces hash values as an intermediary between unique identifiers and user data. Instead of directly using unique identifiers that can track users, the system converts them into hash values that cannot be reversed to obtain original identifiers. This intermediary layer maintains tracking functionality while protecting user privacy by preventing direct identification from the stored values.
Solution Approach 2:
The patent transforms unique identifiers through hashing, changing their parameter representation. The original unique identifier is converted into a hash value with different mathematical properties - it appears random and cannot be inverted to reveal the original identifier. This parameter transformation maintains the uniqueness needed for tracking while eliminating the privacy risk associated with storing actual identifiers.
2Measurement precision
If query results are provided without modification, then data accuracy is improved, but privacy leakage risk increases
Solution Approach 1:
The patent introduces noise values as an intermediary layer between the original query results and the displayed output. These noise values are added to the hash values before presentation, creating a buffer that prevents direct extraction of sensitive information while maintaining the overall statistical accuracy of the results for analytical purposes.
Solution Approach 2:
The patent modifies the parameters of query results by adding controlled noise to hash values. This parameter change ensures that individual records cannot be precisely reverse-engineered from the output, yet the aggregate statistical properties remain intact for legitimate analytical uses, thus balancing accuracy with privacy protection.
3Object-affected harmful factors
If noise values are added to query results, then privacy protection is improved, but data precision deteriorates
Solution Approach 1:
The patent applies local quality by adding noise selectively at the individual record level while maintaining overall statistical integrity. The noise is applied locally to each hash value to prevent identification, but the aggregate properties across multiple records preserve the statistical accuracy needed for analytical purposes, thus protecting privacy without completely sacrificing data utility.
Solution Approach 2:
The patent uses partial action by adding only the minimum necessary noise to achieve privacy protection. Rather than obscuring all data completely, the noise level is calibrated to provide sufficient privacy protection while minimizing the impact on data precision, allowing legitimate analytical uses to continue with acceptable accuracy.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
One embodiment sets forth a method for promoting privacy when providing query results. According to some embodiments, the method can be implemented by a computing device, and includes the steps of (1) receiving at least one query that includes (i) a unique identifier (ID) associated with a software application, (ii) a date, and (iii) at least one parameter, (2) generating query results based on the query, wherein the query results include one or more records, (3) obtaining a random seed value based on the unique ID and the date, (4) generating, for each record of the one or more records, a respective hash value based on (i) the unique ID, (ii) the random seed value, and (iii) at least one aspect of the record, (5) generating respective noise values for the respective hash values, (6) adjusting each record of the one or more records based at least in part on the respective noise value to produce a corresponding adjusted record, and (7) causing information based at least in part on the one or more corresponding adjusted records to be included in a user interface (UI) that is output on a display device.