Loading...
Please wait, while we are loading the content...
Similar Documents
Uma abordagem baseada em hiper-heurística e otimização multi-objetivo para o teste de mutação de ordem superior
| Content Provider | Semantic Scholar |
|---|---|
| Author | Lima, Jackson A. Prado |
| Copyright Year | 2017 |
| Abstract | One of the main testing goals is to determine test sets with a high probability of revealing faults. Mutant Analysis is a promising criterion due to its ability to reveal faults, although with a high computational cost. In order to decrease the mutation testing cost, studies employ the use of Higher Order Mutants (HOMs). The use of HOMs can contribute to decrease the number of equivalent mutants, decrease the test effort and simulate faults close to the real ones. However, the generation of the best HOMs is a complex task, due to the large number of mutants that may exist, and to other factors that influence the generation, such as the efficacy of the generated HOMs. To solve such a problem, some works have successfully applied Search-based Software Engineering techniques through the use of optimization techniques. However, it is still needed to have knowledge about the problem behavior, to determine the best strategy to be applied, and to know how to design and configure the algorithms by choosing the different search operators and defining their parameters in order to improve the search. In this sense, the use of hyper-heuristics allows a more flexible approach to automating these tasks. Also, the use of a hyper-heuristic for selection of different existing strategies to generate HOMs can be useful to reduce the tester’s effort. Considering all these facts, this work proposes a multi-objective approach, called Hyper-Heuristic for Generation of Higher Order Mutants (HG4HOM), which uses the hyper-heuristic concept to generate sets of HOMs. The goal is to select a small number of HOMs which are difficult to kill, and that contribute to improve the test efficacy, that is, it is desired the test cases that kill the selects HOMs are also capable of killing their corresponding FOMs (First Order Mutants). The approach is implemented and evaluated with two multi-objective algorithms: NSGA-II and SPEA2, and three selection methods: Choice-Function (CF), Fitness-Rate-Rank based Multi-Armed Bandit (FRR-MAB ), and random selection (Random). The SPEA2 algorithm using the hyper-heuristic concept together with the CF selection method obtained the best results. In comparison with respect to the traditional strategies, the approach achieved similar results related to the mutation score and statically equivalent values to the best strategy considering the size of the adequate test case sets. The approach obtained the best results when considering the Euclidean Distance values of the solutions with respect to the goals proposed. |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | https://acervodigital.ufpr.br/bitstream/handle/1884/58043/R%20-%20D%20-%20JACKSON%20ANTONIO%20DO%20PRADO%20LIMA.pdf?isAllowed=y&sequence=1 |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
