Model-Based Testing is an approach for testcase generation for black-box testing. From a model specifying the behaviour of the system under test (or a sub-set thereof), test cases are generated, often driven by some notion of coverage of the model. The test cases are then applied to the implementation.

Well-factored models have each information in just one place. For several reasons, this might not be true for the implementation. One reason may be that several people work on different modules and therefore implement the same functionality in both models.

Behaviour Model De-Factoring is a pre-processing step, modifying the model before generating tests from it, with the intention to produce test suites that cover more potential deviations of the implementation compared to the model.

Behaviour model de-factoring uses patterns to split-up factored behaviour that might be factored differently in an implementation, thereby allowing a model-based test case generator to also create tests that catch faults in parts of that behaviour that would be otherwise masked.

As an example, consider a UML state machine. A state may have an entry action and two incoming transitions. In that case one can create a functionally equivalement model where the entry action is duplicated and moved to the transitions.

The concept has been shortly described in [MBT-MDF1]. It can be applied to different variants of Model-Based Testing e.g. for Mutation-Based Test Case Generation.