Xref: utzoo ont.events:1280 uw.talks:2 uw.cs.grad:2
Path: utzoo!utgpu!watmath!thunder!water!wlrush
From: wlrush@water.waterloo.edu (Wenchantress Wench Wendall)
Newsgroups: ont.events,uw.talks,uw.cs.grad
Subject: MASTER'S ESSAY PRESENTATION
Keywords: Mr. David Hsu, graduate student,
Message-ID: <2585@water.waterloo.edu>
Date: 10 Aug 89 17:41:53 GMT
Distribution: ont
Organization: U of Waterloo, Ontario
Lines: 43

Dept. of Computer Science, will speak on 
`` A Processing Model for Redirection Clues in Discourse.''

DEPARTMENT  OF  COMPUTER  SCIENCE  UNIVERSITY OF WATERLOO SEMINAR
ACTIVITIES

MASTER'S ESSAY PRESENTATION

                    -Thursday, August 17, 1989

Mr.   David  Hsu,  graduate  student,  Dept. of Computer Science,
will  speak  on  ``A  Processing  Model  for Redirection Clues in
Discourse.''

TIME:                 11:00 a.m.

ROOM:              DC 3540

ABSTRACT

This  essay examines the processing of interruptions in discourse
based   on   the   model   proposed  by  Grosz and Sidner.  Three
types  of interruptions with redirection clue  words  or  phrases
are  proposed  in  the  Grosz  and  Sidner  model  -  flashbacks,
digressions  and  semantic returns.  We  develop  an interruption
processing  model  (IPM)  and  present  an  implementation, which
takes  as input   discourse   that   includes  interruptions  and
produces   as  output an interpretation of the structure of  that
discourse.   The design of algorithms for the processing model is
guided  by  the  high-level  mechanisms  suggested  by  Grosz and
Sidner.  The  algorithms  present  low-level  descriptions of the
updates  required for the representation  of  the  discourse,  as
stored   in  the  intentional  structure  and  attentional state.
Several  examples   are  provided to illustrate the structures of
discourse  at  any  given  instance.  The  examples  also help to
explain the behavior of clues in discourse.

The  processing  model also forms part of the realization of  the
computational   argument  analysis system by R.  Cohen.  Together
with  the implementation of connective clues   by   T.   Smedley,
they   establish   the   clue  interpreter    in    the   system.
Together   with   the implementation  of  Evidence  Oracle  by M.
Young, they constitute the processing model for the system.