Exploratory Testing For Unwanted Behavior Using Evolutionary Learning Techniques

Abstract

We present a method that allows to create tools that help human testers to do exploratory testing for unwanted behavior of software systems and their components. The general idea is to use evolutionary learning to find interaction schemes with the tested system that result in the tested system showing the unwanted behavior or coming near to it. We present two case studies that instantiate our method and that resulted in detecting not previously known unwanted behavior in both tested systems.