Enriching Reverse Engineering with Annotations

Successful reverse engineering needs to take into account human knowledge about architecture, about features or even about validation of the results of automatic analyses. This knowledge should be linked to the automatically reverse engineered model and should be taken into account by analyses. Typically, when we want to reason about data, we first encode an explicit metamodel and then express analyses at that level. However, human knowledge is often implicit and as a consequence it is not possible to describe it comprehensively upfront. In this dissertation we propose a generic approach to iteratively enrich the system model with external knowledge using annotations. Our mechanism allows the reverse engineer to iteratively describe and refine the annotations during the analysis process, instead of requiring the meta-model to be built upfront. As a validation of the expressiveness of our framework, we show how we use it to support reverse engineering scenarios.