%============================================
%	(C)1992 Institute for New Generation Computer Technology
%	$BG[I[$=$NB>$O(BCOPYRIGHT$B%U%!%$%k$r;2>H$7$F2<$5$$(B
%	(Read COPYRIGHT for detailed information.)
%==========================================
i open the window

diagram is an augmented grammar

the structural relations are holding among constituents

it is not tied to a particular domain of applications

diagram analyzes all of the basic kinds of phrases and sentences

this paper presents an explanatory overview of a large and complex
grammar that is used in a sentence

the annotations provide important information for other parts of the
system that interpret the expression in the context of a dialogue

for every expression it analyzes, diagram provides an annotated
description of the structural relations holding among its constituents

procedures can also assign scores to an analysis, rating some
applications of a rule as probable or as unlikely

the natural language processing system used originally in duals is
implemented in parallel, using layered stream method

a load distribution method is studied for it in order to realize a
very fast natural language parser

the pax analyzes natural language sentences, makes parse trees and
displays them

parsing is performed by bottom up and parsing messages between
processes which correspond to each node of parse tree

the parsing program is generated from definite clause grammar by a
translator, which adds load distribution code automatically

all solutions are searched in the parsing algorithm

however, the result is not always unique because of the ambiguity of
grammar in natural language

the layered stream method which is a broadly applicable paradigm for
all solution search problems is used

a load allocation which minimizes inter pe communication is examined

the pax is a natural language processing system

it analyzes natural language sentences, makes parse trees and displays
them in the window

this system is divided into three parts which are the input part of
sentences, the analysis part which is parser and the output part of
results which are parse trees

the parser is the main part of this system and this part runs on the
multi psi in parallel

the parsing program is generated from definite clause grammar by a
translator

the input program and the output program are written in esp and run on
the front end processor that is psi ii machine

for parsing, a bottom up breadth first algorithm that is called left
corner parsing is used

generally, all solutions are searched in the parsing algorithm

however, the result is not always unique because of the ambiguity of
grammar in natural language

in general, we use the following way to write all solution search
programs in languages that do not support the backtracking mechanism
such as kl1

first, we gather all solutions as a set and the set is gradually
sieved to select the adequate ones

for the pax, the layered stream method that is a broadly applicable
paradigm for all solution search problems is used

the pax is a very fast natural language processing system divided into
three parts which are the input part of sentences, the output part of
results which are parse trees and the analysis part which analyzes
natural language sentences using a bottom up breadth first algorithm
that is called left corner parsing and the layered stream method that
is a broadly applicable paradigm for all solution search problems, and
the input program and the output program run on the front end
processor that is psi ii machine and the parser is the main part of
this system which runs on the multi psi in parallel

this paper provides an explanatory overview of a large and complex
grammar that is used in a sentence, important information for other
parts of the system that interpret the expression in the context of a
dialogue and diagram which provides an annotated description of the
structural relations holding among its constituents for every
expression and analyzes all of the basic kinds of phrases and
sentences and procedures that assign scores to an analysis and rating
some applications of a rule as probable or as unlikely

