Predictive mutation testing method

Abstract

The invention discloses a predictive mutation testing method which is divided into a training stage and a prediction stage. An execution result of each mutant is predicted through a prediction method; the training stage comprises the steps of executing the mutants, conducting characteristic collection on the executed mutants to form a characteristic attribute file and creating a classification model by applying a classification algorithm; the prediction stage comprises the steps of conducting characteristic collection on the unexecuted mutants and classifying each unexecuted mutant by means of the classification model to obtain a classification result. When the predictive mutation testing method is adopted for mutation testing, the mutants to be tested can be predicted by directly using a created prediction model, the execution results of the mutants can be obtained without executing the mutants, and the mutation testing cost can be significantly reduced.

Description

technical field [0001] The invention relates to the technical field of software testing, in particular to a predictive mutation testing method (Predictive Mutation Testing) which can reduce the cost of mutation testing. Background technique [0002] Test cases play an important role in ensuring software quality. Well-designed test cases can find more bugs than poor-quality test cases. In order to measure the error-discovering ability of test cases, mutation testing (mutation testing) came into being, and has been widely concerned and used by more and more researchers and developers in recent years. [0003] Simply put, mutation testing simulates real errors in the program by implanting errors in the program, and then uses existing test cases to find the ability to find implanted errors to predict its ability to find real errors. In mutation testing, the implanted error is called a mutant. When a test case finds that the behavior of a mutant is inconsistent with the behavio...

AI Technical Summary

Problems solved by technology

Selective mutation testing technology is currently a popular technique for reducing the cost of mutation testing. This technology executes a part of the variants in the program and uses the variation scores of the executed variants to predict the mutation score of the overall variant. However, this method In , the execution result of each variant cannot be predicted

Images