Synthetic Test Data Generation via Linked Node Collapse
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data processing software testing methods often rely on randomly generated test data, which fails to accurately replicate the structural patterns and unique aspects of real data, leading to inaccurate test results and potential missed issues when used with actual data sets.
Innovation Solution
A computerized method and system for generating test data based on linked data nodes, where data fields are identified, linked data nodes are created with probability values and data value rules, and parent nodes are collapsed into child nodes to form combined nodes, enabling the creation of test data that accurately reflects real data distributions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If random or pseudo-random data is used for testing, then testing can be performed without real data, but the test data does not exhibit structural patterns of real data leading to inaccurate test results
Solution Approach 1:
The patent segments the test data generation process into multiple independent components: data node creation, rule assignment, probability value assignment, and data generation. Each component handles a specific aspect, making the complex system manageable and modular while producing accurate test data that reflects real data patterns
Solution Approach 2:
The patent uses probability values as parameters to control the generation of test data. By adjusting probability values assigned to different data nodes and rules, the system can replicate the statistical patterns and distributions of real data, thereby improving test result accuracy while maintaining flexible control over data characteristics
2Reliability
If real data is used for testing, then accurate test results can be obtained, but privacy concerns and regulations prevent its use
Solution Approach 1:
The patent creates synthetic copies of real data through a structured generation process that replicates the statistical properties, relationships, and patterns of real data without using actual sensitive information. The system generates test data that mimics real data characteristics while being completely synthetic, thus eliminating privacy concerns and regulatory restrictions
Solution Approach 2:
The patent introduces an intermediary layer of data transformation rules and probability models that mediate between the need for realistic test data and the prohibition against using real sensitive data. This intermediary system generates synthetic data that preserves necessary patterns without exposing actual sensitive information
Data Source
AI summary
The disclosure herein describes generating test data based on a set of linked data nodes and associated data value rules. A plurality of data fields of a data structure is identified. A plurality of linked data nodes is generated based on the identified data fields, wherein each data node comprises a probability value and data value rules for generating data values for the data fields. Parent nodes of the linked data nodes are collapsed into associated child nodes to form combined data nodes. Test data values for the plurality of data fields are generated based on the combined data nodes and a data structure instance is populated with the generated test data values. The populated data structure instance is provided for use in executing tests. The generation of test data based on the data value rules of linked data nodes provides a realistic data environment for testing.