%A D.J. DeWitt
%T A machine independent approach
to the production of optimized horizontal microcode
%R Ph.D. thesis
%I University of Michigan
%D June 1976

%A Stephanie Forrest
%T A study of parallelism in the Classifier System
and its application to classification in KL-ONE semantic networks
%R Ph.D. thesis
%I University of Michigan
%D 1985
%X ABSTRACT:
The dissertation shows how one model of fine-grained parallelism, the
Classifier System, can be used to implement a set of useful operations
for the classification of knowledge in semantic networks.  The
Classifier System appears amenable to hardware implementation, but for
the dissertation, a software simulation was written.  The
"classification" problem was selected as the focus of the
investigation because it is a central problem for many knowledge-
based systems.  Of the various knowledge representation paradigms in
use today, the KL-ONE family has addressed the problem of
classification most directly.  The study therefore focused on the
classification operations of KL-ONE.
The study was divided into three major phases: designing and
implementing the general operations for controlling the Classifier
System, reformulating the KL-ONE formalism in terms of these
operations, and analyzing the efficiency of the parallel algorithms
with respect to the inherent computational tradeoffs among the number
of processors, length of computation, and degree of inter-processor
communication.  The study concludes that architectures of this type
are capable of significantly reducing the time complexity of common
semantic network operations.