Recognizers for extracting architectural features from source code

Architectural representation can play a pivotal role throughout the life cycle of any software program. In particular, we are interested in the role it plays in the maintenance/evolution of legacy programs. During these phases, analysts often describe programs using architectural terminology (e.g., “interfaces”, “interprocess communication”, “layers”, “objects”). Our research and development goals center on supporting such activities through architectural recovery tools that are based on reverse engineering technology. These tools start with existing source code and extract architecture-level descriptions. We have implemented a framework for architectural recovery and our experience leads us to several observations about the representational needs of a library that is populated with families of architecture recognition rules. This paper characterizes the kinds of recognizers we have developed and describes an approach for rule parameterization and retrieval