Potential programming plan and domain concept detection supported by slicing technique

Abstract

Little has been written about the component provision problem (programming plans and domain concepts) for the repositories of knowledge in program understanding systems and reports that specifically address this issue are unknown. Model-driven and plan-driven approaches use repositories where construction and evolution are informal and depend on the experts. In domains where the expert is not available the only valid source of information is the source code. But, domain concept design and programming plans from the source code by non-expert professionals is not a trivial task. In this paper, a hybrid top-down/bottom-up approach based on algorithmic pattern matching and slicing techniques has begun to be defined in order to provide plans or concepts (termed activities in the model-driven approaches) to the knowledge base. Initially, the exploration area is restricted by slicing techniques, and then, software inspection tools are used to further limit the relevant areas. Finally, the initially hypothesized pattern is matched with these segments.