Processing Problems in the IRUS Queries: An Empirical Verification of Problem Coverage in the Natural Language Sourcebook
Patricia Mutch · 1990
Abstract : The Natural Language Sourcebook (Read, Dyer, Baker, Mutch, Butler, Quilici, & Reeves, 1990), a database of processing problems an intelligent computer system might have with natural language, was developed as part of the Artificial Intelligence Measurement System (AIMS). One important use of the Sourcebook is as a means for evaluating intelligent computer systems with regard to their ability to handle specific processing problems. This paper provides an empirical verification of the problem coverage in the Natural Language Sourcebook by referencing output from one intelligent computer system, IRUS (bates, Stallard, and Moser, 1985), to the Sourcebook exemplars.