11. 0644 Workshops

Humanist Discussion Group (humanist@kcl.ac.uk)
Sat, 14 Mar 1998 19:15:05 +0000 (GMT)

Humanist Discussion Group, Vol. 11, No. 644.
Centre for Computing in the Humanities, King's College London

Date: Sat, 14 Mar 1998 13:59:28 -0500 (EST)
From: "David L. Gants" <dgants@english.uga.edu>
Subject: 2 Coling-ACL'98 Workshops

>> From: Priscilla Rasmussen <rasmusse@cs.rutgers.edu>

NEW workshop Call for Papers: Evolutionary Computational Linguistics
and Second CFP: Senseval and the Lexicography Loop




16th August 1998
University of Montreal


There has been a recent resurgence of interest in the application of
evolutionary theory to the understanding of language development,
typology, acquisition and change, and to the development of NLP
systems. Two significant developments potentially distinguish this new
work from that undertaken 100 years or so ago (and ultimately banned
by the French Philological Society). Firstly, a rich body of
mathematical work in population genetics, (non-linear) dynamic
systems, game-theoretic models of evolution, and so forth has since
been developed by theoretical biologists, complexity theorists and
others. Secondly, the study of evolutionary processes has been further
enhanced by the use of computational modelling techniques (in the
simulation of adaptive behaviour, artificial life, etc.) which have
enabled researchers to gain insight into processes too complex for
full mathematical analysis. Evolutionary computation (in the form of
genetic algorithms, genetic programming, hybrids of GAs and neural
networks, etc.) has also been studied and deployed for practical
engineering purposes, including tasks such as grammar induction,
disambiguation, and so forth.

The evolutionary approach is of direct relevance to NLP, and
computational linguists are in a strong position to make a significant
contribution to the development of this research. Synchronic
generative linguistics models a language as a static well-formed
(grammatical) set of strings (sentences) focussing on the (ideal)
individual speaker and her idiolect at a single moment in time. Much
of current NLP technology is based on implementation of generative
models of idiolects and is consequently brittle when it comes into
contact with the reality of language variation and language change
across idiolects and across time. The crucial shift in perspective
provided by the evolutionary approach is to study *changing
populations* of (ideal, generative) speakers. Once this step has been
taken, language is naturally modelled as a dynamic system emergent from
individual idiolects: variation between idiolects interacts with the
process of language learning, leading to `imperfect' or selective
transmission (inheritance) between generations of speakers. This
changes the distribution and composition of idiolects in the
population and thus causes some forms of language change. Once it is
recognised that `bias' in language learning, production and
interpretation creates selection pressure for more learnable,
producible and interpretable variants, then it becomes natural to
treat language as a (complex) adaptive system responding dynamically
to such (often conflicting) pressures.

We are soliciting papers on any aspect of evolutionary computation and
language for this one-day workshop. We hope that the workshop will
stimulate further interest amongst computational linguists and will
provide a forum for cross-fertilisation of ideas between those
applying evolutionary computation to practical NLP tasks and those
using similar techniques to address issues in language acquisition,
change and variation.


Bob Berwick (MIT, USA) berwick@ai.mit.edu.
Ted Briscoe (Cambridge Univ, UK) ejb@cl.cam.ac.uk.


Steve Abney ATT Research
Jim Hurford Edinburgh University
Bill Keller Sussex University
Partha Niyogi Bell Labs
Luc Steels Sony, Paris


Submissions should be full length papers between 3500--5000 words
on A4/US letter in 11/12pt Times Roman or similar font. Preferably, email
self-contained latex source to the co-chairs using
http://coling-acl98.iro.umontreal.ca/colaclsub.sty, or send 5
hardcopies to the address below.

Ted Briscoe
Computer Laboratory
University of Cambridge
Pembroke St.


Submission Deadline: April 20, 1998
Notification Date: June 1, 1998
Camera ready copy due: June 22, 1998


* *

Sponsored by ACL SIGLEX and EURALEX

There are now many automatic Word Sense Disambiguation (WSD) programs
but it is currently very hard to determine which are better, which
worse, and where the strengths and weaknesses of each lie. There is
widespread agreement that the field urgently needs an evaluation
framework. Under the auspices of ACL SIGLEX and EURALEX, a pilot will
take place in the course of 1998. As in ARPA evaluation exercises,
the framework comprises:

1) definition of task and scoring metric
2) preparation of a set of manually tagged correct answers
3) a dry run, with sample data distributed to participants
4) distribution of test data to participants;
participants sense-tag and return;
taggings scored against correct answers
5) workshop to discuss results, lessons learned, way forward

We shall be undertaking evaluation for at least English, French,
Italian and Spanish. The workshop will be held at

Herstmonceux Castle, Sussex, UK
Sept 2-4 1998

If you have a working WSD program (or will have one by Summer 1998),
and would like to subject it to objective, quantitative evaluation, or
if you have skills or resources that you would like to contribute to
the exercise, first look at


and then mail your expression of interest to



As soon as possible:
expressions of interest to be registered
By end of:
April 98: 'dry run' data samples distributed to participants
June 98: test data distributed to participants, and, one week
later, returned with taggings
July 98: correct answers and scores made available
Sept 98: workshop

The workshop will comprise:

(1) reports/papers on the manual tagging and other aspects of
the evaluation exercise
(2) reports/papers from participants on the design and
performance of their system
(3) other related research papers
(4) working sessions on the way ahead for WSD evaluation

The deadline for "other related research papers" is

16th April 1998

* Preference will be given to papers discussing WSD, with particular
attention to evaluation issues
* Maximum submission length: 6 pages
* First page to include title, abstract, and author's name(s) and
contact details
* Electronic submission of postscript documents permitted
but must be supported by hard copy to arrive not later than 23rd April
(in case of printing problems).

email: senseval-submissions@itri.bton.ac.uk
hard copy: SENSEVAL Submissions
University of Brighton
Lewes Road
Brighton BN2 4GJ, UK

Full details available at http://www.itri.bton.ac.uk/events/senseval/

Adam Kilgarriff
SENSEVAL co-ordinator

Humanist Discussion Group
Information at <http://www.kcl.ac.uk/humanities/cch/humanist/>