Identifying error proneness in path strata with genetic algorithms

Abstract

In earlier work we have demonstrated that GA can successfully identify error prone paths that have been weighted according to our weighting scheme. In this paper we investigate whether the depth of strata in the software affects the performance of the GA. Our experiments show that the GA performance changes throughout the paths. It performs better in the upper, less in the middle and best in the lower layer of the paths. Although various methods have been applied for detecting and reducing errors in software, little research has been done into partitioning a system into smaller, error prone domains for software quality assurance. To identify error proneness in software paths is important because by identifying them, they can be given priority in code inspections or testing. Our experiments observe to what extent the GA identifies errors seeded into paths using several error seeding strategies. We have compared our GA performance with random path selection.