The software architecture description is often the reasoning basis for important design decisions. Nevertheless, efforts to keep an architecture description up-to-date through the software evolution life cycle are challenging. Rigorous evaluation must be taken on the as-implemented architecture description to identify design violations. However, the architecture description involves dealing with a broad variety of concerns, making it unfeasible to evaluate it in its entirety. The state of the art approaches for evaluating software architectures have put little emphasis on the extraction of architectural information with respect to a set of communication patterns and quality characteristics of interest. In our previous works, we presented ARAMIS, an approach to support the evaluation and reconstruction of software architecture with a strong emphasis on software behavior. This master thesis presents the extension of ARAMIS for enabling the specification of views, viewpoints, and perspectives, which can be applied on architecture descriptions to retrieve a more focused and relevant view of software architectures. Furthermore, to facilitate the process of identifying the best evolution variant, the developed methodology enables the simulation of software architecture evolution. The evaluation has shown very promising results.