Using genetic programming to evolve action selection rules in traversal-based automated software testing: results obtained with the TESTAR tool

摘要: Traversal-based automated software testing involves an application via its graphical user interface (GUI) and thereby taking the user’s point of view executing actions in a human-like manner. These are decided on fly, as under test (SUT) is being run, opposed to set up form sequence prior testing, that then used exercise SUT. In practice, random choice commonly decide which action execute at each state (a procedure referred monkey testing), but number alternative mechanisms have also been proposed literature. Here we propose using genetic programming (GP) evolve such selection strategy, defined list IF-THEN rules. Genetic has proved be suited for evolving all sorts programs, rules particular, provided adequate primitives (functions terminals) defined. must aim extract most relevant information from SUT dynamics process. We introduce problem hand evaluate their usefulness based various metrics. carry out experiments compare results with those obtained by Q-learning, reinforcement learning technique. Three applications Software Under Test experiments. The analysis shows potential GP strategies.