A metrics-driven method for assessing reverse engineering tool capabilities used to reverse legacy source code components to design component abstractions

The research presented in this dissertation studies and develops metrics-driven methods and supporting tools to reverse source code components to design components. Such methods are comprised of two phases. In the first phase a metrics-checklist is developed to help determine which legacy source code components should be kept for reversing to design components or alternatively should be discarded. In the second phase a metrics-checklist is developed to help assess reverse-engineering tool capabilities. Based on the assessment in the second phase an appropriate set of reverse-engineering tools may be selected for use in deriving design components from source code components. The thesis is that the metrics-driven method is more reliable than and at least as efficient as an alternative non-metrics, driven assessment method. In this thesis it is asserted that the metrics-driven method is more reliable than the alternative method since the metrics-driven method will uncover more defects when assessing a tool's ability to reverse legacy source code components to design components. It is asserted that the metrics-driven method is at least as efficient as the alternative method since it will take the same amount of effort or less effort to assess the reverse-engineering tools. A conceptual model and experimental model were developed and used to help justify these assertions. In developing part of the conceptual model, existing tools assessment methods were analyzed to determine how they could benefit and be improved by use of a metrics-driven method. Using the experimental model, an experiment was conducted to test the above two assertions. Documented findings supported these two assertions.