Mutation 2020

Scientists have created many cost reduction tech-niques for mutation testing, and most of them reduce cost withminor losses of effectiveness. However, many of these techniquesare difficult to generalize, difficult to scale, or both. Publishedresults are usually limited to a modest collection of programs.Therefore, an open question is whether the results of a given costreduction technique on programs studied in the paper will holdtrue for other programs. This paper introduces a conceptualframework, namedSiMut, to support the cost reduction ofmutation testing based on historical data and program similarity.Given a new, untested programu, the central idea is applyingtouthe same cost reduction strategy applied to a groupGofprograms that are similar touand have already been tested withmutation, and check for consistency of results in terms of reducedcosts and quality of test sets.SiMutincludes activities to computeprogram abstractions and similarity. Based on this information,it supports the application of mutation cost reduction techniquesto bothGandu. This paper presents the concepts behindSiMut,a proof-of-concept implementation ofSiMut, and results from apilot study. Finally, we discuss some issues related to the use ofSiMut, focusing on the composition of a representative dataset toproperly explore the potential of our framework.