Classifying mutants with decomposition kernel

Abstract

The paper deals with the problem of reducing the cost of mutation testing using artificial intelligence methods. The presented approach is based on the similarity of mutants. The mutants are coded as control flow diagrams representing the programs structure, variables and conditions. The similarity is then calculated with the use of a new graph kernel and used to predict if a given test case detects a mutant or not. The prediction process is performed by a classification algorithm. Experimental results are also presented in this paper on the basis of two systems.