[ont.events] A Definite Clause Version of Categorial Grammar.

ylfink@water.waterloo.edu (ylfink) (05/07/88)

DEPARTMENT OF COMPUTER SCIENCE
UNIVERSITY OF WATERLOO
SEMINAR ACTIVITIES

ARTIFICIAL INTELLIGENCE SEMINAR

                    - Thursday, May 12, 1988

Dr.  Remo  Pareschi,  University of Pennsylvania,  will
speak  on  ``A  Definite  Clause  Version of Categorial
Grammar''.

TIME:                11:00 AM

ROOM:              DC 1302

ABSTRACT

Categorial  Grammar is a framework for natural language
analysis  where  grammatical  constituents  are  viewed
either  as  functions  or arguments, and are associated
with  corresponding  syntactic  types.   We  present  a
first-order  version  of such a framework, based on the
idea  of  encoding syntactic types as definite clauses.
Under this approach, there are no explicit requirements
of   adjacency  between  combinable  constituents,  and
word-order  constraints are obtained simply by allowing
subformulae of complex types to share variables ranging
over  string  positions.  It is in this way possible to
give a straightforward account of problematic syntactic
constructions   involving  discontinuous  constituents,
which are particularly common in natural language.  The
second half of this enterprise consists of showing how,
for   this   formalism,   parsing  can  be  efficiently
implemented  as  theorem  proving.   Our  approach   to
encoding   types   as  definite  clauses  presupposes a
modification   of   standard   Horn   logic  syntax  to
allow    internal  implications  in  definite  clauses.
This   modification  is needed to account for the types
of   higher-order  functions  and,  as  a  consequence,
standard  Prolog-like Horn logic theorem proving is not
powerful enough. We tackle this problem by adopting  an
intuitionistic   treatment   of  implication, which has
been  proposed  elsewhere  as   a   useful extension of
Prolog  for  implementing  hypothetical  reasoning  and
modular   logic programming.  We then provide a parsing
algorithm  in terms of a decision procedure which tries
to  assign  a  type to a given input string,  where the
size   of   the   logic  program  needed  for  a   type
assignment  is always linear with respect to the length
of the input.