An Evolutionary Mutation Testing System for Java Programs: eMuJava

Abstract

Mutation-based testing is costly but we can bring it down with combination of evolutionary testing approaches. Search-based mutation testing combines mutation and evolutionary testing to exploit the advantages offered by both of these techniques. The evolutionary techniques like genetic algorithm supports automating the test case generation during software testing that can reduce a lot of testing resources. The processes of mutation testing can also be automated to further save the testing cost. In this research, we present a testing tool, eMuJava that can perform mutation-based testing of Java-based programs automatically. eMuJava is fully automated system that performs all the activities of mutation testing and test case generation automatically. For mutation testing, it supports conventional as well as object-oriented mutation operators. eMuJava is implementing four testing techniques including three different genetic algorithms and random testing. The system offers complete control to the tester to perform testing of Java programs, allows monitoring of all the steps that tool performs, and all intermediary outputs that the tool generates. We have evaluated the tool by performing extensive experiments on Java programs. The outcomes of an experiment include test case set, mutation score, and statistical information about the experiment. We have also statistically evaluated the experimental results to prove their effectiveness. The generated test cases are further analyzed on different set of mutation operators to evaluate their strength.