Predication of Program Behaviours for Functionality Testing
Jin Hu, Yi Wang, Nian-Wei Chen, Shuo Wang, Ling-Ming Zeng · 2009
Functionality Testing is one of the most popular used technologies of black-box testing strategy. Programs to be test were taken as implicit functions mapping inputs into outs. In the view of functionality testing, if the functional behaviors of the program can be predicated correctly, it could be applied as test oracles for newer test cases generation. The main problems of this issue are discussed as: firstly, collecting test data for the software being tested on the base of covering its functionalities. Secondly, training Self-Organized-Feature-Mapping networks with the attained data set to simulate and predicate program behaviors. Thirdly, generating test oracles with the pre-described networks that is able to predicate program behaviors with unknown inputs automatically. Lastly, designing and doing experiments to validate the effectiveness of such method. This article concludes that by simulating the program behaviors with properly designed ANN can provide valuable clues for automatic functionality testing.