Integrating cognitive support with CASE-tools for design recovery

Abstract

Reverse engineering (RE) activities account for the largest part of current expenses in software maintenance. The RE support provided by existing design tools is limited to simple mappings of idioms in the source code to diagrammatic primitives. Human analysts still have to go through the laborious task of manually detecting patterns and creating higher abstractions. Perhaps the most important challenge in automating RE is to deal with the imperfect knowledge inherently involved in the detection process. Recently, a number of researchers have developed prototypes of design tools with knowledge-based RE capabilities. For several reasons these research prototypes are rarely acceptable for industrial-strength applications. Consequently, innovative technologies often have difficulties reaching their target audience. We try to address this issue by adopting established design tools and extending them with knowledge-based RE functionality. This paper reports on the development of such an extension component and contains a case study that shows the feasibility of this approach.