From genetic to bacteriological algorithms for mutation‐based testing

摘要: SUMMARY The level of confidence in a software component is often linked to the quality its test cases. This can turn be evaluated with mutation analysis: faults are injected into (making mutants it) check proportion detected (‘killed’) by But while generation set basic cases easy, improving may require prohibitive effort. paper focuses on issue automating optimization. application genetic algorithms would appear an interesting way tackling it. optimization problem modelled as follows: case considered predator mutant program analogous prey. aim selection process generate able kill many possible, starting from initial predators, which provided programmer. To overcome disappointing experimentation results, .Net components and unit Eiffel classes, slight variation this idea studied, no longer at ‘animal’ (lions killing zebras, say) but bacteriological level. indeed better reflects issue: it mainly differs one introduction memorization function suppression crossover operator. purpose explain how have been adapted fit resulting algorithm so much that has given another name: algorithm. Copyright c