A Novel Regression Test Selection Method with Graph-Based Genetic Algorithm

Authors : Ramazan OZKAN, Zeynep ORMAN, Ruya SAMLI

Pages : 1-12

View : 31 | Download : 24

Publication Date : 2023-08-10

Article Type : Research Paper

Abstract :Regression testing is a re-running test to ensure that previously developed and tested software is not affected by changes. Testing a software after changes is important and necessary in order to maintain the software development and maintenance processes. However, repeating all tests after each change is not feasible especially in large-scale projects. Regression test selection which means selecting a subset of tests has emerged as a solution to this issue. This paper presents a GBGA insert ignore into journalissuearticles values(graph-based genetic algorithm); with the most compatible neighbor crossover as a solution to the regression test selection problem. In this GBGA, each individual in the population is located on a node of predefined graph structure and the probabilities of the crossover are limited depending on the neighborhood relations to increase population diversity, prevent premature convergence, and refine the convergence performance. This GBGA is applied to this problem to find the minimum set of test cases to enhance the performance of the GA by locating populations on graphs and limiting the crossover option with neighborhood connections to increase the diversity. The results show that the proposed GBGA with the most compatible neighbor crossover has superior performance in terms of fitness value when compared to genetic algorithm.
Keywords : Regression test selection, graph based genetic algorithm, compatible crossover, optimization