Institute of Information Science, Academia Sinica

esting is used to provide confidence in the correctness of software.

A test suite that detects a lot of faults may be considered to be effective, but it could just be that the program under test is very faulty. Mutation testing provides a measure of test suite quality using artificial faults. As we know the number of faults introduced, we can precisely determine the proportion revealed by a test suite. We say that test suites are effective if they identify a difference in the output of many mutants. On the other hand, mutants are considered effective if it is difficult to distinguish them from the original program. How can we reconcile these two criteria? Search-based software engineering has the answer.

In this talk, I will provide a brief introduction to mutation testing and then explain how search based software engineering may be used to make it more effective. I will describe a method for optimising test strategies to achieve a higher mutation score with fewer test cases, a simple heuristic for mutant similarity and a technique for adjusting the similarity of mutants parametrically. The aim of this talk is to encourage an exchange of research ideas between Academia Sinica and the University of York. For this reason, my wife will give an overview of the Department of Computer Science at York and there will be plenty of time for questions!