Hierarchical Clustering Data Anonymization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data anonymization methods, such as k-anonymity and perturbation, face challenges in high-dimensional datasets, where achieving privacy while maintaining statistical information is difficult, especially due to the complexity of achieving k-anonymity and the limitations of l-diversity and t-closeness, and the destruction of correlations in data through noise addition.
Innovation Solution
A system and method using hierarchical data clustering and perturbation, where high-dimensional data is converted into a normalized vector space, clustered, and then perturbed using methods like 'assign' or 'shuffle' to achieve k-anonymity while preserving statistical properties, utilizing techniques like k-nearest neighbor and attribute-based division.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If generalization is used to achieve k-anonymity, then privacy protection is improved, but computational complexity increases and statistical information is lost
Solution Approach 1:
The patent segments the data anonymization process into two distinct phases: (1) hierarchical clustering to organize and group similar records, and (2) perturbation to anonymize within clusters. This segmentation allows the system to achieve k-anonymity through perturbation rather than generalization, reducing computational complexity while preserving statistical information.
Solution Approach 2:
The patent changes the approach from attribute-level generalization to record-level perturbation. By applying perturbation techniques (such as swapping records within clusters) rather than generalizing attributes, the system achieves k-anonymity while maintaining the original statistical properties and correlations of the data.
2Reliability
If perturbation is applied to anonymize data, then privacy protection is improved, but correlations among attributes are destroyed
Solution Approach 1:
The patent segments records into clusters based on their similarity and statistical properties before applying perturbation. This ensures that perturbation occurs within homogeneous groups, preserving attribute correlations while still achieving privacy protection through k-anonymity.
Solution Approach 2:
The patent applies different perturbation strategies to different clusters based on their local characteristics. By tailoring the perturbation to each cluster's specific structure and statistical properties, the system maintains correlations within local groups while achieving overall privacy protection.
3Reliability
If attribute suppression is used to remove identifiers, then privacy protection is improved, but re-identification risk increases through linking with external datasets
Solution Approach 1:
The patent creates synthetic copies of records within clusters through perturbation. By swapping and redistributing attribute values within clusters to achieve k-anonymity, the system creates indistinguishable copies that prevent re-identification while maintaining the statistical structure of the original data.
Data Source
AI summary
A system and method for data anonymization using hierarchical data clustering and perturbation is provided. The system includes a computer system and an anonymization program executed by the computer system. The system converts the data of a high-dimensional dataset to a normalized vector space and applies clustering and perturbation techniques to anonymize the data. The conversion results in each record of the dataset being converted into a normalized vector that can be compared to other vectors. The vectors are divided into disjointed, small-sized clusters using hierarchical clustering processes. Multi-level clustering can be performed using suitable algorithms at different clustering levels. The records within each cluster are then perturbed such that the statistical properties of the clusters remain unchanged.