AI Test Case Selection Using RNN Confidence and Defect Prediction

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing software testing techniques require manual intervention, leading to delays and inaccuracies in selecting relevant test cases for new software testing requirements, affecting the software development cycle.

Innovation Solution

An automated method using a Recurrent Neural Network (RNN) model to determine confidence scores for existing test cases and a linear regression model to predict defect slippage rates, allowing for the selection of relevant test cases based on predetermined thresholds.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If manual techniques are used for mapping existing test cases with new requirements, then flexibility and adaptability are maintained, but delay and inaccuracy are introduced in software testing

Engineering Contradiction:

Improveaccuracy in selecting test casesVSAvoiddelay in software testing

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent replaces manual mapping techniques with an automated machine learning system that uses Recurrent Neural Networks to compare new requirements with existing test cases. This substitution eliminates human intervention in the matching process, thereby removing delays and inaccuracies associated with manual operations while maintaining high precision through algorithmic comparison.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system enables test case selection to be performed autonomously by the machine learning model without requiring human operators. The RNN model automatically processes new requirements, compares them against the existing test case pool, and identifies relevant test cases independently, making the system self-sufficient and eliminating time loss from manual involvement.

Inventive Principle:

Principle #25Self-service

2Productivity

If manual intervention is used in test case selection, then complex judgment can be applied, but the software product development cycle is affected due to delay and inaccuracy

Engineering Contradiction:

Improvespeed of software testingVSAvoidaccuracy in test case selection

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The patent replaces manual judgment processes with an automated machine learning system that uses Recurrent Neural Networks to compare new requirements with existing test cases. This substitution eliminates human intervention in the matching process, thereby removing delays and inaccuracies associated with manual operations while maintaining high precision through algorithmic comparison.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Measurement precision

If a large pool of test cases is maintained for comprehensive coverage, then testing thoroughness is improved, but the difficulty of selecting relevant test cases increases

Engineering Contradiction:

Improverelevance identification accuracyVSAvoidcomplexity of test case selection process

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The machine learning model acts as an intermediary between the large pool of existing test cases and the new requirements. The RNN model processes the requirements and automatically identifies relevant test cases from the extensive pool, simplifying the selection process by introducing an intelligent intermediary that handles the complexity of matching large datasets.

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentUS11327873B2System and method for identification of appropriate test cases using artificial intelligence for software testing

Publication Date: 2022.05.10 WIPRO LTD

AI summary

A method and a system of selecting test cases from existing test cases for a new software testing requirement are disclosed. In an embodiment, the method may include determining a confidence score associated with each of existing test cases, based on comparing a new software testing requirement with the existing test cases using a Recurrent Neural Network (RNN) model, and selecting a set of test cases from the existing test cases based on the confidence score and a predetermined threshold confidence score. The method may further include predicting a defect slippage rate associated with each test case using a linear regression model, and shortlisting a sub-set of test cases from the set of test cases based on the predicted defect slippage rate associated with each test case and a predetermined threshold defect slippage rate.