Generation of Hypertext User Documentation from Usage Models

The certification of software reliability requires a testing approach that closely matches the intended software usage. This requirement is fulfilled by statistical testing through an underlying usage model which describes all possible interactions with the software as well as their probability. Usage models allow to generate test cases automatically, saving time in the test design phase. Nevertheless, the effort to design a usage model can not be neglected. In order to gain additional benefits from their design, this diploma thesis defines an extended usage model that allows to generate user documentation. In this way, the application of usage models becomes more attractive for software development projects. The extended usage model is based on a finite state machine formal- ism as in conventional usage model design. Additionally, it describes 1) user interface elements and how to use them, and 2) scenarios on how to solve tasks with the help of t he software in terms of consecutive u ser actions. A prototype has been implemented in Java that parses a usage model description file and generates HTML user documentation. So far, the prototype is capable of generating a tutorial section of a user documen- tation which describes how to solve typical tasks. An example of an extended usage model based on a small application is introduced to evaluate the prototype and to elicit potentials of the approach taken. The example shows that it is possible to generate readable, structured hypertext documentation based on an extended usage model.