A Collaborative Environment for Authoring Large Knowledge Bases

  • Authors:

  • Peter D. Karp;Vinay K. Chaudhri;Suzanne M. Paley

  • Affiliations:

  • Pangea Systems, 4040 Campbell Avenue, Menlo Park, CA 94025. pkarp@pangeasystems.com;Artificial Intelligence Center, SRI International, 333 Raenswood Avenue, Menlo Park, CA 94025. vinay.chaudhri@sri.com;Pangea Systems, 4040 Campbell Avenue, Menlo Park, CA 94025. paley@pangeasystems.com

  • Venue:
  • Journal of Intelligent Information Systems
  • Year:
  • 1999

Quantified Score

Hi-index 0.00

Visualization

Abstract

Collaborative knowledge base (KB) authoring environmentsare critical for the construction of high-performance KBs. Suchenvironments must support rapid construction of KBs by a collaborativeeffort of teams of knowledge engineers through reuse of existingknowledge and software components. They should support themanipulation of knowledge by diverse problem-solving engines even ifthat knowledge is encoded in different languages and by differentresearchers. They should support large KBs and provide a scalable andinteroperable development infrastructure. In this paper, we presentan environment that satisfies many of these goals.We present an architecture for scalable frame representation systems(FRSs). The Generic Frame Protocol (GFP) provides infrastructure forreuse of software components. It is a procedural interface to framerepresentation systems that provides a common means of accessing andmodifying frame KBs. The Generic KB Editor (GKB-EDITOR) providesgraphical KB browsing, editing, and comprehension services for largeKBs. Scalability of loading and saving time is provided by a storagesystem (PERK) which submerges a database management system in an FRS. Multi-user access is controlled through a collaboration subsystem thatuses a novel optimistic concurrency control algorithm. All the results have been implemented and tested in thedevelopment of several real KBs.