Anonymized Dataset Linking via Equivalence Class Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems face inefficiencies in linking anonymized datasets with non-anonymized datasets due to computing inefficiencies, particularly in executing join operations, which are slow and not supported by all database systems, making it costly and time-consuming to evaluate re-identification risk across large datasets.
Innovation Solution
The solution involves transforming the traditional join operation by identifying equivalence classes and applying a generalization level to link anonymized datasets with non-anonymized datasets, using a processor to efficiently match records across various types of anonymization operations, supporting both global and local recoding approaches, and handling large datasets with billions of records.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional join operations are used to link anonymized datasets with non-anonymized datasets, then data integrity and accuracy are maintained, but computing speed and efficiency deteriorate significantly
Solution Approach 1:
The patent segments the anonymized dataset into multiple equivalence classes based on generalized attributes. Each equivalence class groups records with identical generalized values, allowing the system to process and link records in batches rather than individually. This segmentation enables parallel processing and reduces the computational complexity of join operations while maintaining data integrity through the preservation of equivalence class relationships.
Solution Approach 2:
The patent performs preliminary generalization of attributes in the anonymized dataset before executing join operations. By pre-computing equivalence classes and generalized attribute values, the system prepares the data in advance to enable faster linking with non-anonymized datasets. This preliminary action reduces the computational burden during the actual join operation, significantly improving computing speed without sacrificing data accuracy.
2Measurement precision
If traditional join operations are used to link anonymized datasets, then accurate record matching is achieved, but processing time and computational cost increase
Solution Approach 1:
The patent divides the dataset into equivalence classes based on generalized attribute values, allowing the system to process and match records in grouped batches rather than individual comparisons. This segmentation reduces the number of join operations required while maintaining matching accuracy through the preservation of equivalence class relationships, significantly reducing processing time for large datasets.
Solution Approach 2:
The patent changes the parameter representation by generalizing specific attribute values into equivalence class identifiers. Instead of comparing detailed attribute values during join operations, the system uses generalized equivalence class labels that preserve matching semantics while reducing computational complexity. This parameter transformation maintains record matching accuracy while dramatically reducing processing time.
3Productivity
If equivalence classes and generalization levels are used to speed up linking, then computing efficiency improves, but system complexity increases
Solution Approach 1:
The patent implements a universal equivalence class framework that works across different types of anonymization operations and dataset structures. The same generalization and equivalence class mechanisms can be applied to various attribute types and anonymization methods, providing a multi-functional solution that improves computing efficiency without requiring separate complex systems for different scenarios. This universality reduces overall system complexity despite the advanced techniques employed.
Data Source
AI summary
Various embodiments are provided for linking of anonymized datasets in a computing environment are provided. A number of linking records may be identified between an anonymized dataset and one or more non-anonymized datasets of a knowledge base according to one or more equivalence classes and a generalization level.