Research And Application On The Subsumption Of C Mutant

Mutation testing is a fault-based technique to assess test adequacy by creating a set of artificial faults,called mutants,and calculating how many of them are detected(or killed)by tests.An existing problem is that there are many tedious mutants which contribute nothing to analysis but cost raised.Recently,mutation testing researchers develop a graph model to describe redundancy among mutations by subsumption.The subsumption can identify minimal mutants among mutant set to reduce testing cost and improve the quality of fault detection.Unfortunately,it is undecidable to determine the subsumption between two given mutants.This paper used a fast algorithm which is an approximate method to identify subsumption and minimal mutants with fewer testing times and computations.This paper's contribution contains:1)Propose a fast algorithm to construct dynamic mutant subsumption graph by clustering,ranking and pruning;2)Support identifying minimal mutants on 77 C mutants on the basis of the Code Testing System;3)Apply the fast algorithm to identifying minimal mutants in 12 programs about 60 thousand mutants including 5 programs in Siemens Suite.The empirical study shows that the algorithm succeeds in reducing 25.6%comparisons between mutants,respective to exhaustive approach.The result also proves that most mutants are redundant with only about 2.15%of mutants is minimal mutants.Through the algorithm,we can identify minimal mutants efficiently as to support analysis for mutation testing.