Ontology based negative selection approach for mutation testing

Abstract

Mutation testing is used to design new software tests and evaluate the quality of existing software tests. It works by seeding faults in the software program, which are called mutants. Test cases are executed on these mutants to determine if they are killed or remain alive. They remain alive because some of the mutants are syntactically different from the original, but are semantically the same. This makes it difficult for them to be identified by the test suites. Such mutants are called equivalent mutants. Many approaches have been developed by researchers to discover equivalent mutant but the results are not satisfactory. This research developed an ontology based negative selection algorithm (NSA), designed for anomalies detection and similar pattern recognition with two-class classification problem domains, either self (normal) or non-self (anomaly). In this research, an ontology was used to remove redundancies in test suites before undergoing detection process. During the process, NSA was used to detect the equivalent mutant among the test suites. Those who passed the condition set would be added to the equivalent coverage. The results were compared with previous works, and showed that the implementation of NSA in equivalent mutation testing had minimized local optimization problem in detector convergence (number of detectors) and time complexity (execution time). The findings had more equivalent mutants with average of 91.84% and scored higher mutation score (MS) with average of 80% for all the tested programs. Furthermore, the NSA had used a minimum number of detectors for higher detection of equivalent mutants with the average of 78% for all the tested programs. These results proved that the ontology based negative selection algorithm had achieved its goals to minimize local optimization problem.