Mutated Specification-Based Test Data Generation with a Genetic Algorithm

Abstract

Specification-based testing methods generate test data without the knowledge of the structure of the program. However, the quality of these test data are not well ensured to detect bugs when non-functional changes are introduced to the program. To generate test data effectively, we propose a new method that combines formal specifications with the genetic algorithm (GA). In this method, formal specifications are reformed by GA in order to be used to generate input values that can kill as many mutants of the target program as possible. Two classic examples are presented to demonstrate how the method works. The result shows that the proposed method can help effectively generate test cases to kill the program mutants, which contributes to the further maintenance of software.

Performance of Enhanced Multiple-Searching Genetic Algorithm for Test Case Generation in Software Testing

“…The search-based technique is widely used for test case generation [19][20][21][22]. The following subsections describe some of the most well-known search-based techniques before introducing the proposed EMSGA.…”

Multiple-Searching Genetic Algorithm for Whole Test Suites

“…GA is used to improve test case generation for multiple paths, for which GA can find solutions quickly [26]. GA is used to reform formal specifications to generate test data that can detect as many faults as possible [27]. The evidence accumulated suggested that GAs enhance the performance of test cases, allowing them to examine more source code and detect more faults.…”

Automatic Group Organization for Collaborative Learning Applying Genetic Algorithm Techniques and the Big Five Model

“…Genetic algorithms were described by Holland [30] and are considered as a computational model family inspired by evolutionary biology [31]. To date, their use has spread beyond the original conception, as a more general type of evolutionary algorithm that attempts to simulate Darwinian evolution and natural selection through the recombination and mutation of individuals [32].…”

“…In general, a genetic algorithm is structured in a three-step iterative process [31]: (i) an initial population of solutions (individuals) is created, represented by a chromosome that encodes the solution to the problem; (ii) a group of individuals is selected through a specific strategy, based on the fitness function, and the next population is generated by applying genetic operators (crossover and mutation) to the selected individuals; (iii) step (ii) is repeated until the remaining individuals in the generation are good enough according to the fitness function and the stop criteria. This process is outlined in Figure 1.…”