Topic-driven testing.

When manually testing Web sites humans can go with vague, yet general instructions, such as \"add the product to shopping cart and proceed to checkout\". Can we teach a robot to follow such instructions as well? In this paper I present a novel model, called semantic usage patterns which allows us to capture the general topics behind the individual steps of interactions. These models can be extracted from existing test descriptions be they in natural language or in form of system tests. Those usage patterns can be applied even on applications they were not designed for. They allow to test applications automatically in order to identify behavioral anomalies in the application model or detect missing functionalities.