The definition of oracles - mechanisms (program, process or data) used to judge the correctness of a program under test - is key to successful automation of software testing activities. The automation of oracles is not a simple task and this activity becomes even more costly when the decision regarding the correctness of the system must take into account complex outputs, such as, a file that represents a sound message, a graphical interface or a Virtual Reality environment. When reactive or real time systems such as speech synthesis programs are cosidered the complexity becomes even greater. The literature is scarce on flexible strategies for automating oracles for these domains. One possible source of information for oracle automation may be the feature extraction of sound signals, images or virtual scenes produced by the systems under test. These characteristics serve as a source of information to indicate the correctness or otherwise of applications. This project aims at the implementation and the generation of resources to the automatization of testing oracles for systems whose output is given in complex form. The implementation of plugins for the framework O-FIm (Oracle for Image), developed in previous work, enables the automation of oracles for different test domains such as graphical interfaces and Web applications. O- FIm uses concepts of CBIR (Content-Based Image Retrieval) as a strategy to automated oracles to particular domains. As a result, the work is expected to contribute to the Software Engineering through empirical evaluations and experiments concerning the applicability of this strategy in systems whose output is given in complex formats. Furthermore, the implementation of plugins for the tool opens the possibility of technology transfer between academia and industry.
News published in Agência FAPESP Newsletter about the scholarship: