Logical Data Masking via Clustering and Shuffling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data masking methods are inadequate for securely masking sensitive customer information, particularly in handling both numerical and alphabetical data, and often result in poor data integrity and inefficiencies, such as difficulty in designing test cases, complexity in special cases, and ineffective processing of small data sets.
Innovation Solution
A method and system for logical data masking using clustering algorithms to group and shuffle numerical data, combined with string masking algorithms for alphabetical data, date masking, and unique code masking, ensuring robust data integrity by minimizing the possibility of identical or similar entries post-masking.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If traditional data masking methods (nulling out, character masking, substitution, shuffling) are used, then data security is improved, but data integrity and usability deteriorate
Solution Approach 1:
The patent applies parameter changes by transforming numerical data through clustering algorithms that group entries based on statistical parameters (mean, variance, skewness, kurtosis) and then applying transformations to these parameters. This allows the data to maintain statistical properties while being masked, resolving the contradiction between security and integrity.
Solution Approach 2:
The patent creates masked data entries that copy the structural and statistical properties of original data entries while replacing actual values. The masked data maintains the same distribution patterns, data types, and relational structures, enabling test cases to be designed effectively while preserving security.
2Object-affected harmful factors
If shuffling algorithms are used for data masking, then data security is improved, but effectiveness deteriorates for small data sets
Solution Approach 1:
For small data sets, the patent uses parameter-based clustering that analyzes statistical characteristics (mean, variance, skewness, kurtosis) to group entries. This parameter-driven approach is more effective than random shuffling for small data sets because it leverages the limited statistical information available to create meaningful groups, improving both security and processing effectiveness.
3Object-affected harmful factors
If multiple masking algorithms are applied to different data types, then data security is improved, but system complexity increases
Solution Approach 1:
The patent implements a universal data masking system that handles multiple data types (numerical, alphabetical, dates, unique codes) through a single integrated framework. The system uses a common clustering algorithm that can process different data types by transforming them into appropriate representations, eliminating the need for separate masking systems for each data type and reducing overall complexity.
Solution Approach 2:
The patent segments the masking process into distinct stages: data classification by type, clustering based on statistical parameters, and application of type-specific masking algorithms. This segmentation allows each component to be optimized independently while maintaining overall system coherence, reducing complexity through modular design.
4Object-affected harmful factors
If clustering algorithms group numerical data entries, then data security is improved, but computational time increases
Solution Approach 1:
The patent uses parameter-based clustering that computes statistical parameters (mean, variance, skewness, kurtosis) to group data entries. By using these summary statistics rather than examining all data points individually, the algorithm achieves effective clustering with reduced computational time, making it suitable for large data sets while maintaining security.
Data Source
AI summary
A system and method for logically masking data by implementing masking algorithms is provided. The method includes receiving one or more inputs from user regarding type of data masking to be implemented depending on type of data entry. Data entries include alphabetical data, data comprising unique codes, data comprising dates and numerical data. Based on inputs received, the data entries are classified and appropriate masking algorithms are executed. For masking numerical data entries, the data entries are first grouped using clustering algorithms and are then shuffled using shuffling algorithms. For low level of data masking selected by a user, numerical data entries are shuffled within groups and for high level of data masking selected by a user, numerical data entries are shuffled across groups.