A Production System Version of the Hearsay-II Speech Understanding System
CARNEGIE-MELLON UNIV PITTSBURGH PA DEPT OF COMPUTER SCIENCE
Pagination or Media Count:
A prime candidate organization for large, knowledge-rich systems is that of a production system PS. PSs are rule-based architectures that have been used successfully for tasks ranging from models of human behavior to large application systems in chemistry and medicine, to classical artificial intelligence programs. The question studied by this thesis is whether a PS architecture PSA helps or hinders with respect to implementation problems encountered by Hearsay-IIHSII, a large artificial intelligence system for understanding speech, developed at Carnegie-Mellon University CMU. This is an important question because many of these problems, such as efficiency, compensating for error, controlling directionality, augmenting knowledge, and analyzing performance, have become limiting factors for performance. To obtain an answer to this question, an actual system called HSP, for HearSay-Production system was implemented on C.mmp, the CMU multi-miniprocessor, with a portion of th HSII speech knowledge translated into productions. An early decision was made to maintain close comparability of HSP with HSII rather than explore the more general question of how to best understand speech with a PS. Two knowledge- source KS programs from a complete HSII configuration were completely translated and run in HSP, and these provide a basis for some detailed comparisons between HSII and HSP. Ten other KSs were translated, and their static structure provides supporting evidence.