The use of description logics in KBSE systems

Abstract

The increasing size and complexity of many software systems demand a greater emphasis on capturing and maintaining knowledge at many different levels within the software development process. This knowledge includes descriptions of the hardware and software components and their behavior, external and internal design specifications, and support for system testing. The Knowledge-based software engineering (KBSE) research paradigm is concerned with systems that use formally represented knowledge, with associated inference precedures, to support the various subactivities of software development. As they growing scale, KBSE systems must balance expressivity and inferential power with the real demands of knowledge base construction, maintenance, performance, and comprehensibility. Description logics (DLs) possess several features—a terminological orientation, a formal semantics, and efficient reasoning procedures—which offer an effective tradeoff of these factors. We discuss three KBSE systems in which DLs capture some of the requisite knowledge needed to support design, coding, and testing activities. We then survey some alternative approaches (to DLs) in KBSE systems. We close with a discussion of the benefits of DLs and ways to address some of their limitations.