ABSTRACT In the reverse engineering of a software program, one of the key difficulties is actually to understand the software. While the published techniques work top down or bottom up, our approach works middle-out: before trying to understand the low level code, we first rebuild a hypothetical analysis model from the use cases of the system. This model then represents the target of the understanding task. In fact we try to map the code elements to the analysis objects. For this approach to be useable in large industrial software systems, it must be supported by a powerful tool. This paper presents the Eclipse plug-in we developed to support our methodology, as well as a reverse engineering scenario using this tool. We then discuss the technology we used and the result we obtained.
