ML-Based Test Coverage Prediction for Automated Traceability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current code testing methods require manual connection of new requirements to test sets and lack validation, and defects are often isolated to specific test instances without considering broader impacts on other requirements.
Innovation Solution
A system and method using machine learning to predict test coverage for new application requirements, creating new test features automatically and determining defect severity by analyzing affected test features and requirements, enabling 100% test coverage and automatic traceability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual connection of new requirements to test sets is performed, then traceability is established, but time consumption and labor effort increase significantly
Solution Approach 1:
The patent replaces the manual mechanical process of connecting requirements to test sets with an automated machine learning-based system. The ML model automatically analyzes new requirements, predicts relevant test features, and establishes traceability connections without human intervention, thereby eliminating time loss while maintaining validation reliability.
Solution Approach 2:
The system enables self-service automation where the machine learning model independently performs the entire traceability establishment process. The model automatically processes requirements, identifies relevant test features, creates connections, and validates traceability without requiring manual human operation, thus resolving the time consumption issue.
2Measurement precision
If defects are connected only to specific test instances, then direct causality is identified, but broader impact on other requirements is missed
Solution Approach 1:
The patent implements a multi-functional traceability system where the machine learning model not only connects defects to specific test instances but also automatically identifies and connects affected requirements. This universal approach allows the system to perform both precise defect localization and broader impact analysis simultaneously, preventing information loss about affected requirements.
Solution Approach 2:
The system establishes bidirectional feedback loops where defect detection triggers automated analysis of affected test features and requirements. The ML model uses this feedback to update traceability connections and identify ripple effects across the system, ensuring both precise measurement and comprehensive information retention about defect impacts.
3Productivity
If existing test features are reused for new requirements, then testing efficiency improves, but complete test coverage may not be achieved
Solution Approach 1:
The patent applies partial action by having the machine learning model selectively reuse existing test features only for the portions of new requirements that they cover, while automatically generating additional test features for uncovered portions. This approach maintains high productivity through reuse while ensuring complete coverage by adding necessary new tests.
Solution Approach 2:
The system segments the test coverage analysis into two parts: covered portions handled by existing test features and uncovered portions requiring new test features. The ML model automatically divides the new requirement analysis and creates appropriate test features for each segment, thereby maintaining efficiency while achieving completeness.
Data Source
AI summary
As described herein, a system, method, and computer program are provided for test-related automatic tracing. A new requirement defined for an application is identified. The new requirement is processed, using a machine learning model, to predict each portion of the new requirement covered by existing test features. One or more new test features are caused to be created for the new requirement, based on the prediction of each portion of the new requirement covered by existing test features.