Automated Software Architecture Analysis and Visualization Advanced of Large, Mixed-Language Systems

Large, complex software systems are difficult to maintain, modify, and enhance. Information regarding the overall system architecture and individual component design is often sparse, and usually informal. In order to effectively address legacy system maintenance and evolution, we must provide facilities to recognize and represent the information abstractions which exist throughout a software system, at every design layer. Our objective is to investigate the feasibility of leveraging a combination of: a) language-neutral models, b) source code analysis techniques, and c) user interaction with graphical representations, to extract and maintain higher-level software knowledge. Our fundamental innovation is this combination of analysis results -- to propose and highlight potential abstractions, and cooperative user interaction -- to accept, elaborate, or modify them, all supported by an underlying suite of extensible source code analysis and exploration tools.