Automatically testing interactive applications using extended task trees

Task trees are common notations used to describe the interaction between a user and an interactive application. They contain valuable information about the expected user behaviour as well on the expected software reactions and, thus, they can be used to support model-based testing. In this paper, a method for automatically generating test data from task trees is introduced. The task tree notation is extended to support operational profile specification. The user behaviour is automatically extracted from such extended trees as a probabilistic finite input-output state machine, thanks to formal semantics defined for this purpose for the task tree operators. The resulting probabilistic machine can then be used to generate test data simulating the user behaviour. This simulation can be performed using Lutess, a testing environment developed for synchronous software. The translation of the user interaction model into a Lutess description is explained and experimental results are reported. (C) 2009 Elsevier Inc. All rights reserved.