Reference: Fikes, R. & Farquhar, A. Large-Scale Repositories of Highly Expressive Reusable Knowledge. Knowledge Systems Laboratory, April, 1997.
Abstract: We present a vision of next generation tools and services that will enable the widespread development and use of computer interpretable ontologies. Central to that vision is the notion of distributed ontology repositories resident on multiple ontology servers containing large-scale highly structured ontologies and supported by sophisticated ontology construction, testing, merging, extraction, reformulation, and translation tools. The key enabler in the distributed ontology repository architecture is a network application programming interface (API) for ontology servers that will support storage, manipulation, and access to the contents of ontologies on a server. We describe how OKBC, an API specifically designed to provide knowledge-level interoperability among server and client systems, provides that support. We then consider the criteria for an ontology representation language and an agenda of extensions to current ontology representation languages that address major deficiencies in those languages and appear to be attainable in next generation languages. Finally, we address the issue of what reasoning is needed to support ontology repository construction and use, and describe a deductive retrieval facility under development for the Ontolingua ontology server that includes a theorem prover which runs as a background task to reformulate sentences so that are accessible by the server's special purpose high speed retrieval methods.
Notes: (Updated March 1998).
Full paper available as ps.