A Novel Metaheuristic Based Method for Software Mutation Test Using the Discretized and Modified Forrest Optimization Algorithm

Arasteh B., Gharehchopogh F. S., Gunes P., Kiani F., Torkamanian-Afshar M.

Journal of Electronic Testing: Theory and Applications (JETTA), vol.39, no.3, pp.347-370, 2023 (SCI-Expanded) identifier

  • Publication Type: Article / Article
  • Volume: 39 Issue: 3
  • Publication Date: 2023
  • Doi Number: 10.1007/s10836-023-06070-x
  • Journal Name: Journal of Electronic Testing: Theory and Applications (JETTA)
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, Aerospace Database, Communication Abstracts, Compendex, INSPEC, Metadex, DIALNET, Civil Engineering Abstracts
  • Page Numbers: pp.347-370
  • Keywords: Bug-prone codes, Forest optimization algorithm, Mutation score, Software mutation test
  • Yıldız Technical University Affiliated: Yes


The number of detected bugs by software test data determines the efficacy of the test data. One of the most important topics in software engineering is software mutation testing, which is used to evaluate the efficiency of software test methods. The syntactical modifications are made to the program source code to make buggy (mutated) programs, and then the resulting mutants (buggy programs) along with the original programs are executed with the test data. Mutation testing has several drawbacks, one of which is its high computational cost. Higher execution time of mutation tests is a challenging problem in the software engineering field. The major goal of this work is to reduce the time and cost of mutation testing. Mutants are inserted in each instruction of a program using typical mutation procedures and tools. Meanwhile, in a real-world program, the likelihood of a bug occurrence in the simple and non-bug-prone sections of a program is quite low. According to the 80–20 rule, 80 percent of a program's bugs are discovered in 20% of its fault-prone code. The first stage of the suggested solution uses a discretized and modified version of the Forrest optimization algorithm to identify the program's most bug-prone paths; the second stage injects mutants just in the identified bug-prone instructions and data. In the second step, the mutation operators are only injected into the identified instructions and data that are bug-prone. Studies on standard benchmark programs have shown that the proposed method reduces about 27.63% of the created mutants when compared to existing techniques. If the number of produced mutants is decreased, the cost of mutation testing will also decrease. The proposed method is independent of the platform and testing tool. The results of the experiments confirm that the use of the proposed method in each testing tool such as Mujava, Muclipse, Jester, and Jumble makes a considerable mutant reduction.