Anonymization Boundary Alignment for Accurate Cluster Merging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing anonymization processes face challenges in maintaining accuracy of analysis after merging clusters with low similarity, leading to decreased analysis precision due to varying cluster anonymization methods across organizations.
Innovation Solution
An anonymization apparatus and method that utilize boundary information to divide and integrate data clusters, ensuring k-anonymity by aligning cluster granularity and minimizing information loss through methods like Mondrian partitioning and cluster integration/deletion based on consented data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If clusters are merged based on similar average values, then merging can be performed, but the similarity of clusters as a whole remains low due to varying anonymization methods across organizations
Solution Approach 1:
The patent segments the anonymization process into two distinct phases: (1) a determination apparatus that merges data and generates boundary information indicating cluster boundaries, and (2) anonymization apparatuses that use this boundary information to divide their respective data into clusters. This segmentation ensures that all organizations use consistent cluster boundaries derived from merged data, resolving the contradiction between merging capability and cluster similarity.
Solution Approach 2:
The determination apparatus performs preliminary anonymization and cluster boundary determination on merged data before distributing boundary information to individual anonymization apparatuses. This preliminary action establishes a unified reference framework that guides subsequent local anonymization processes, ensuring that clusters across different organizations align in terms of boundaries and similarity while maintaining各自的数据 privacy.
2Adaptability or versatility
If anonymization processes vary by organization, then local data processing is flexible, but the accuracy of analysis after merging decreases
Solution Approach 1:
The patent implements local quality by allowing each anonymization apparatus to process its own data using the boundary information received from the determination apparatus. Each organization's anonymization apparatus divides its data into clusters based on the predetermined boundaries, maintaining local processing flexibility while ensuring that the cluster structures are consistent across organizations, thereby preserving analysis accuracy after merging.
3Productivity
If clusters are divided without unified boundary information, then processing is independent, but cluster granularity alignment is poor leading to information loss
Solution Approach 1:
The boundary information generated by the determination apparatus serves as an intermediary that bridges the independent processing of different anonymization apparatuses. This boundary information acts as a common reference framework that enables independent local processing while ensuring that clusters from different organizations align in granularity and structure, thereby minimizing information loss during the merging process.
Data Source
AI summary
An anonymization apparatus includes: an acquiring unit that acquires boundary information indicating the boundary between clusters, specified by performing an anonymization process on predetermined data; and a dividing unit that generates an anonymized cluster set by dividing data possessed by the anonymization apparatus into a plurality of clusters based on the boundary information acquired by the acquiring unit.