Information Processing Device for Data Anonymization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing anonymization methods, such as k-anonymization, face challenges in balancing data security and usefulness, as they often reduce the identifying possibility but lower the data's utility by altering attribute values, making it difficult to maintain statistical integrity and increasing the risk of data leakage.
Innovation Solution
An information processing device determines a replacement attribute and replaces its values based on a useful index and replacement limitations, ensuring both security and data usefulness by selecting an optimum replacement pattern that maintains statistical characteristics while preventing identification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If k-anonymization is performed by converting attribute values to broader categories (e.g., changing specific ages to age groups), then security is improved by reducing identifying possibility, but data usefulness deteriorates because statistical accuracy is lost
Solution Approach 1:
The patent creates synthetic copies of data records by generating artificial objects that replicate the statistical characteristics of the original data. Instead of modifying existing records (which loses information), the system generates multiple synthetic copies that preserve the original attribute values while distributing the identifying information across multiple synthetic records. This allows the original data to remain intact and useful while security is improved through the presence of synthetic copies.
Solution Approach 2:
The patent changes the parameter of data representation by introducing synthetic objects with artificially generated attribute values. Rather than changing the parameters of original data (e.g., converting ages to age groups), the system generates new records with parameter values that statistically match the original distribution. This preserves the usefulness of original data while improving security through the added synthetic records.
2Reliability
If directly identifiable information is deleted from the database, then security is improved by preventing direct identification, but data usefulness deteriorates because the data becomes less complete
Solution Approach 1:
Instead of deleting identifiable information from original records, the patent creates synthetic copies that contain plausible but artificial identifiable attributes. The original records with their complete and accurate information are preserved, while the synthetic copies provide the security layer by diluting the identifying power of any single record through the presence of multiple similar-looking synthetic records.
3Reliability
If many data records are processed to achieve k-anonymization (ensuring k or more records have the same attribute value), then security is improved, but processing complexity and time increase significantly
Solution Approach 1:
The patent performs preliminary action by pre-generating synthetic data records with attribute values that are designed to match the statistical characteristics of the original data. Rather than processing existing records to achieve k-anonymization (which requires examining and modifying many records), the system pre-creates synthetic copies that automatically provide the required anonymity level when added to the dataset.
Solution Approach 2:
The system creates synthetic copies of data records through a copying process that replicates the statistical structure of the original data. This copying approach is more efficient than the alternative of processing and modifying existing records, as it generates new records that inherently satisfy anonymity requirements without requiring complex processing of the original dataset.
Data Source
AI summary
The present invention is configured to determine a replacement attribute as a replacement target of attribute values, from among a plurality of attributes of data including the attribute values of the plurality of attributes corresponding to each of a plurality of objects, and to replace the attribute values of the replacement attribute determined in the data, based on a useful index being an index of usefulness of the data and a replacement limitation being a limitation in replacement of the attribute values.