doi:10.3850/978-981-08-7304-2_0009

ABSTRACT

As software develops and evolves, new test cases are continually generated to validate the atest modifications. Thus the need for automatic testing arises and it suggests the use of model-based testing technology. Among various modeling languages, UML is widely spread and used for its implicity, understandability and ease of use. UML state diagrams are used for specifying the dynamic behavior of the system. Thus many techniques have been developed for automatic generation of test cases from state diagrams. However, a percentage of test cases may become redundant, because of the redundant paths in state diagrams and test cases are generated for all the paths which increase the size of the test suite. Due to the constraints on time and resources, it may be impossible to rerun all test cases whenever the software is modified. Therefore it is desirable to keep test suite sizes manageable by removing redundant test cases. In this paper, a novel approach is presented for reduction of automatically generated test cases from UML state diagrams using graph mining, numerical methods and clustering techniques.

© 2011 ICCA 2010 Organizers