Towards Interactive Model Mining From Embedded Software

The idea of model mining is the extraction of higherlevel models from code. For example, one could extract state machines that describe the behavior of a program. Such models can be very helpful for software maintenance tasks such as program understanding. The major drawback of fully-automatic model extraction is that, when applied on real world systems (e. g., software-intensive systems), the resulting models are complex, difficult, and contain information on a quite low level. Developers can hardly find and understand the information they need. In this paper, we present our ideas towards a framework that solves this problem. The main idea is to let developers contribute to the process of model extraction. The ultimate goal of the framework is to provide high quality representative types of models on an adequate level of abstraction and with low manual effort.