High fidelity knowledge systems

Abstract

Knowledge systems are permeating into practical use, but they can only pay off if users can easily use and understand them. Most contemporary knowledge systems emphasize only the semantic correctness of representations and inference mechanisms. We believe that an increasingly important, but not yet widely appreciated, class of knowledge systems is that of high fidelity knowledge systems whose representation and processes closely resemble the objects and work processes that people now already use. Such systems are much more likely to be accepted into real use, and offer a high potential for reaping the benefits of knowledge system technology. In this paper, we descibe Hypertelligence, an environment for building high fidelity knowledge systems. Hypertelligence makes use of two converging technologies: hypertext and object-based artificial intelligence. We also illustrate the idea of high fidelity and the potential of Hypertelligence with an implemented system for selecting construction scheduling programs. © 1994.