Model-based testing is a process of generating tests for a test target based on a behavioral model. In many cases, suitable models do not exist and building these from scratch is a costly effort. A potential approach to bootstrap this process is to generate models with reverse-engineering methods. Potential sources of information for such models include user sessions, existing test cases, and field data. All of these can be seen as different forms of test data. However, using existing test cases as a basis to create a model for test generation raises the question of what is the benefit over the previously existing tests. This paper aims to answer this question by evaluating the benefits in terms of a practical case study.
展开▼
