Algoritmo meta-heurístico para o problema de geração de dados de teste

Abstract

Search-Based Software Engineering (SBSE) is a research area that integrates Software Engineering and Optimization, in the sense that optimization techniques are applied to deal more efficiently with several recurring problems associated with the software development process. As an example, the literature presents works in the area of SBSE in solving problems in software testing (automation of test data generation), software design(balancing quality of service objectives - QoS, selection, and integration between components ), requirements (determining the set of requirements that balance the costs of software development and customer satisfaction), project management (project planning and resource allocation) and refactoring (automation of refactoring processes). A relevant part of SBSE’s efforts is devoted to software testing problems, particularly test data generation automation. The software tests aim to verify, detect, and correct the programmers possible errors and represent about 50% of the total development cost. The automation of the generation of test data consists of using optimization techniques (in general, meta-heuristics) to determine the set of data that will be used as input of the algorithm in the testing phase, so that code coverage is maximized. That is, the data set generated should be able to test as much of the code under test as possible. Therefore, the objective of the work in question is to develop a metaheuristic approach to solve the problem of automating the generation of test data. One possible approach would be Computational Transgenetics, which was never applied to solve the specified problem. After elaborating the algorithm, we will perform computational experiments to evaluate the developed algorithm.