Supporting the Extraction of Timed Properties for Passive Testing by Using Probabilistic User Models

Testing is one of the most widely used techniques to increase the quality and reliability of complex software systems. In this paper we extend our previous work on passive testing with invariants to incorporate (probabilistic) knowledge obtained from users of the system under test. In order to apply our technique, we need to obtain a set of invariants compiling the relevant properties of the system under test, and this is a time-intensive task. We present a novel approach to extract invariants from a specification, based on the idea that an invariant is better than another one if it can be checked more times in a given log. We present a formal approach where probabilistic user models are incorporated.