Component-Based Support for Building Knowledge-Acquisition Systems

Mark A. Musen, Ray W. Fergerson, William E. Grosso, Natalya F. Noy, Monica Crubezy, and John H. GennariStanford Medical InformaticsStanford UniversityStanford, California 94305-5479U.S.A.Musen@Stanford.EDUhttp://www.smi.stanford.edu/projects/protegeDuring the past decade, there has been increasingconsensus within the knowledge-based–systemscommunity on appropriate conceptual componentsfor building intelligent computer programs. Intelli-gent systems are now generally construed in termsof both domain ontologies and abstract problem-solving methods that operate on knowledge basesdefined in terms of those ontologies. There hasbeen less consensus, however, regarding how tooptimize the operational components and the userinterfaces of tools that assist developers in the con-struction of knowledge-based systems. For the mostpart, such lack of consensus is to be expected, giventhe way in which domain considerations oftendominate the way in which knowledge can best beentered, browsed, and updated in any computer-based tool. In our research group at Stanford Uni-versity, we acknowledge both the central importanceand the great variability of domain-specific idiomsthat can enhance the functionality of knowledge-acquisition tools. Our work seeks specific ways toharness this variability and to allow developers totake advantage of alternative approaches.Our latest development in a series of computer-based knowledge-acquisition systems is known asProtege-2000. Protege-2000 is a meta-tool” that“helps users to construct domain-specific knowledge-acquisition systems that application experts can useto enter and browse the content knowledge of elec-tronic knowledge bases. The architecture of Pro-tege-2000 assumes that, as with Web browsers,users will want to enhance andcustom tailor the system’s behavior by means of a variety of “plug-ins.” These plug-ins are modular pieces of programcode that add new functionalities to Protege-2000 inwell circumscribed ways. Developers can contrib-ute new Protege-2000 plug-ins to a library main-tained on the Internet, and can freely download newplug-ins to augment the behavior of their ownknowledge-acquisition systems constructed usingProtege-2000. Our approach establishes a new kindof knowledge-acquisition enterprise—one of build-ing knowledge-acquisition–tool