Estratégias para controlar o tamanho da suíte de teste gerada a partir de abordagens MBT.

Abstract

Testing is the most commonly applied technique to evaluate the quality of software as part of verification & validation processes. However, it is usually an expensive activity. Promising to reduce costs as well as promoting effectiveness, Model-based Testing (MBT) approaches have been proposed, where test cases can be obtained from specifications. In MBT, the algorithms used to obtain test cases are usually based on a “search” in a behavioral model and, in most of the times, the stop decision is based on structural coverage criteria that are exhaustively applied. Therefore, in this context, the number of applicable test cases tends to be very high. On the other hand, usually, there are not sufficient resources (time and money) to execute all of them. Also, some test cases may exercise common sequences of functionalities. In this sense, redundancy is an important concept that can be considered to obtain a smaller test suite, once that redundant parts may not increase functionality coverage or fault detection. Some strategies for controlling the size of the test suites have been proposed: test case selection and test suite reduction. The former usually considers a test purpose (to reduce a space search) and/or fix a number of test cases that are desired without taking into account the redundancy concept. On the other hand, some strategies for test suite reduction are proposed and experimented considering structural redundancy for white-box testing. Obviously, it is necessary to seek strategies for controlling the size of the test suites generated from MBT approaches that consider the redundancy concept. Different strategies for controlling the size of test suites are proposed in this thesis focusing on selection and reduction. Results show that strategies for selection and reduction based in Similarities are good to detect faults and provide a adequate coverage. Even though the strategies proposed can be applied to different testing levels, the focus is on system testing. Finally, a new way to evaluate test suite reduction strategies - by considering the rate of fault detection - is proposed. Even though, the rate of fault detection is a metric widely used to compare test suite prioritization strategies, it has not yet been considered to evaluate test suite reduction strategies.