LINGUIST List 2.841

Thu 05 Dec 1991

Calls: Text linguistics, CLS

Editor for this issue: <>


  1. , CFP
  2. Eric Schiller, Call for Papers: Chicago Linguistic Society

Message 1: CFP

Date: Tue, 26 Nov 91 10:47:42 EST
From: <weir-5PRC.Unisys.COM>
Subject: CFP
 The AAAI Workshop on Statistically-Based NLP Techniques
 held at the Tenth National Conference on Artificial Intelligence
Interest in statistically-based NLP techniques has grown considerably
over the last five years, partly because of disenchantment with the
rate of technological progress in developing NLP systems within a
strictly ``knowledge-based'' framework. Such systems have suffered
from three chronic problems. First, their reliance upon domain
restrictions tends to result in a lack of robustness when confronted
with gaps in coverage. Second, because domain knowledge is handcoded
in such systems, extending them to support new domains tends to be a
laborious process. Third, such systems generally must be maintained
by developers, not by users.
There are, of course, good reasons why researchers have developed NLP
systems within a knowledge-based framework---some information is very
difficult to capture and represent by statistical means. Rather than
completely abandoning a knowledge-based framework, researchers have
begun to develop hybrid systems in which an effort is made to maximize
the potential of statistically-based and knowledge-based techniques.
With the growing interest in statistically-based techniques, it is
time for a forum on their use in NLP applications. What components of
an NLP system can benefit from such techniques? What tradeoffs exist
in using statistical techniques, and in combining them with
handcrafted knowledge? Are there interesting interactions that arise
when more than one such technique is used? And finally, is there
evidence that a given technique is capable of supporting large-scale
applications---for example, is it reasonable to expect grammar
induction systems to be capable of generating broad-coverage grammars
capable of supporting large-scale data extraction applications, and if
so, are there any special benefits of using such an approach in a
large-scale system?
The objective of this workshop is to establish the capabilities of
existing statistically-based NLP techniques, and to envision how they
may be improved. Discussions of significant success stories and
interesting failures in efforts to employ such techniques within
large-scale NLP applications will be emphasized. Reports on the use
of statistically-based methods in syntactic and semantic analysis will
be especially encouraged, along with reports on efforts to automate
the acquisition of linguistic knowledge from large text and spoken
language corpora.
TOPICS: corpus-based acquisition of linguistic knowledge
 dealing with sparse data
 stochastic grammars
 grammar induction
 statistically-based semantic interpretion
 generation of corpora for statistically-based NLP methods
 formal evaluation of statistically-based techniques
 part-of-speech tagging
FORMAT: The papers that are accepted will determine the format of
 the workshop. If general topics emerge, then the papers
 will be organized into panel sessions with time allocated
 at the end of each session for a brief discussion of the
 session's theme.
ATTENDANCE: 25-50 participants will be invited to present papers
 at the workshop, depending on the quality of the submissions.
 A maximum of two authors per paper will be invited.
SUBMISSION REQUIREMENTS: Five copies of a 10-page abstract must be submitted
 to the address given below.
SUBMIT TO: Carl Weir
 Center for Advanced Information Technology
 Unisys Defense Systems, Inc.
 70 E. Swedesford Rd.
 PO BOX 517
 Paoli, PA 19301
 Center for Advanced Information Technology
 Unisys Defense Systems, Inc.
 70 E. Swedesford Rd.
 PO BOX 517
 Paoli, PA 19301
 Phone: (215) 648-2369
 Fax: (215) 648-2288
 Stephen Abney
 Ralph Grishman
 Courant Institute of Mathematical Sciences, NYU
 Ralph Weischedel
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue

Message 2: Call for Papers: Chicago Linguistic Society

Date: Wed, 4 Dec 91 17:32:19 CST
From: Eric Schiller <>
Subject: Call for Papers: Chicago Linguistic Society
Call for Papers:
 28th Annual Regional Meeting of the
 Chicago Linguistic Society
 April 23-25, 1992
 Main Session (April 23-24)
We invite original, unpublished work on any topic of general
linguistic interest. Invited Speakers are:
 Susumu Kuno, Harvard University
 Charles Kisseberth, University of Illinois at Urbana-Champaign
 Arnold Zwicky, The Ohio State University
 Parasession (April 24-25) on
 The Role of the Cycle in Linguistic Theory
We invite original, unpublished work which supports or refutes the
notion of the cycle in the analysis of language. Abstracts from all
areas of linguistic investigation are welcome, including, but not
limited to, syntax, phonology, and morphology. Invited Speakers
 John Goldsmith, University of Chicago
 James McCawley, University of Chicago
 Geoffrey Pullum, University of California, Santa Cruz
 Jerrold Sadock, University of Chicago
 Elizabeth Selkirk, University of Massachusetts, Amherst
Abstracts: Please submit ten copies of a one-page, 500-word
anonymous abstract (for a 25-minute paper), along with a 3x5 card
with your name, address, phone number, e-mail address, title of
paper, and indication of whether the paper is intended for the main
session or the parasession. The abstract should clearly indicate
the data covered, outline the arguments presented, and include any
broader implications of the work. If necessary, append a page of
data and/or references. An individual may submit at most one single
and one coauthored abstract. Deadline for receipt of abstracts is
February 14th, 1992. Send abstracts to Chicago Linguistic Society,
1010 East 59th Street, Chicago, IL 60637.
For more information: 312-702-8529
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue