Database Anonymization via Column-Specific Generalization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing anonymization techniques for medical databases are inefficient and do not support high-speed processing, failing to effectively anonymize privacy information while adhering to disclosure rules.
Innovation Solution
A database management system that generates and applies anonymization methods by reading only specific columns of a relation table, using generalization rules to create aggregate results that satisfy disclosure rules, thereby enabling fast anonymization processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If existing anonymization techniques are applied to medical databases, then privacy information can be anonymized satisfying disclosure rules, but the processing speed is slow and efficiency is poor
Solution Approach 1:
The patent segments the anonymization process into distinct phases: (1) reading only necessary columns from the relation table, (2) generating anonymization methods for each column separately using generalization rules, (3) aggregating results column by column, and (4) combining column results into the final anonymized table. This segmentation allows the system to process data in manageable units, improving both speed and resource utilization while maintaining anonymization quality.
Solution Approach 2:
The patent applies preliminary action by pre-defining generalization rules for each column before processing the actual data. The system prepares anonymization methods in advance based on column characteristics and disclosure rules, then applies these pre-prepared methods during execution. This preliminary preparation significantly reduces processing time during actual anonymization operations.
2Reliability
If all columns of the relation table are read for anonymization processing, then complete anonymization can be achieved, but processing time and computational resources increase
Solution Approach 1:
The patent extracts and processes only the necessary columns from the relation table based on anonymization requirements and disclosure rules. Instead of reading the entire table, the system identifies and extracts specific columns that contain privacy information requiring anonymization, processes them separately, and combines the results. This extraction approach dramatically reduces processing time and resource consumption while maintaining complete anonymization coverage for relevant data.
3Reliability
If generalization rules are applied to all attribute values, then disclosure rules are satisfied, but the complexity of processing increases
Solution Approach 1:
The patent applies local quality by customizing generalization rules for each specific column based on its characteristics and the type of privacy information it contains. Instead of applying a uniform anonymization approach to all columns, the system defines column-specific generalization rules that are tailored to the local requirements of each attribute. This localized approach simplifies processing by avoiding unnecessary complexity while ensuring disclosure rule compliance for each column type.
Data Source
AI summary
The database management system (DBMS) receives a first instruction specifying anonymization rule information corresponding to a column of the relation table, among anonymization rule information that is present for each column included in the relation table and shows a plurality of generalization rules. The DBMS reads the column from the relation table in response to the first instruction, and generates a temporary result obtained by generalizing each attribute value of the column based on any of a plurality of generalization rules. The DBMS generates an aggregate result obtained by aggregating the temporary result. The DBMS generates an anonymization method including generalization information indicating a correspondence relationship of each attribute value of the column and any of the plurality of generalization rules when the aggregate result satisfies a disclosure rule. The DBMS generates anonymization information as a result of processing the relation table based on the first anonymization method.