Semantic networks constitute one of the many attempts to capture human knowledge in an abstraction suitable for processing by computer program. While semantic nets enjoy widespread popularity, they seem never to live up to their authors' expectations of expressive power and ease of construction. This paper examines the fundamentals of network notation, in order to understand why the “formalism” has not been the panacea it was once hoped to be. We focus here on “concepts”—what net-authors think they are, and how network nodes might represent them. The simplistic view of concept nodes as representing extensional sets is examined, and found wanting in several respects.
A general system to simulate human cognitive processes is described. The four-part system comprises a nodespace to store the network structure ; a supervisor; a transition network parser; and an interpreter. The method by which noun phrases operate and the process f or the determiner "the" is presented. An analysis of verb structures illustrates how network structures can be constructed from primitiv e verb definitions that get at the underlying structures of particular verbs. The paper concludes with an illustratio n of a problem in question-asking.In IJCAI-73: THIRD INTERNATIONAL JOINT CONFERENCE ON ARTIFICIAL INTELLIGENCE, 20-23 August 1973, Stanford University Stanford, California.