[Corpora-List] CoNLL-X: First Call for Papers

Erik Tjong Kim Sang erikt at science.uva.nl
Wed Dec 7 12:01:00 CET 2005

The Tenth Conference on Computational Natural Language Learning
New York City, June 8-9, 2006
First Call for Papers

CoNLL is the yearly conference organized by SIGNLL (the ACL Special
Interest Group on Natural Language Learning). Previous CoNLL meetings
were held in Madrid (1997), Sydney (1998), Bergen (1999), Lisbon
(2000), Toulouse (2001), Taipei (2002), Edmonton (2003), Boston
(2004), and Ann Arbor (2005). This year, CoNLL will be collocated with
HLT-NAACL in New York City.

See http://staff.science.uva.nl/~erikt/signll/ and
http://staff.science.uva.nl/~erikt/signll/conll/ for more
information about SIGNLL and CoNLL. The official Web of CoNLL-X
can be found at http://www.cnts.ua.ac.be/conll/

CoNLL is an international conference for research on natural language
learning. We invite submission of papers about natural language
learning topics, including, but not limited to:

* Computational models of human language acquisition
* Computational models of the evolution of language
* Machine learning methods applied to natural language
processing tasks (speech processing, phonology, morphology,
syntax, semantics, discourse processing, language
engineering applications)
* Statistical methods (Bayesian learning, graphical models,
kernel methods, statistical models for structured problems)
* Symbolic learning methods (rule induction and decision
tree learning, lazy learning, inductive logic programming,
analytical learning, transformation-based error-driven learning)
* Biologically-inspired methods (Neural Networks, Evolutionary Computing)
* Reinforcement learning
* Active learning, ensemble methods, meta-learning
* Learning architectures for structural and relational NLP tasks
* Computational learning theory analysis of language learning
* Empirical and theoretical comparisons of language learning methods
* Models of induction and analogy in linguistics

Special Topic of Interest
Apart from the topics listed above, this year we wish to encourage the
submission of papers that propose learning theories, architectures,
algorithms, methods, or techniques for improving the robustness of
learning-based NLP systems.

One important type of brittleness in current learning-based NLP
systems is domain dependence. Since learning is mainly performed in a
supervised setting, even slight differences between training corpora
and test corpora (text genre, style, new vocabulary, etc.) may cause
substantial degradation in the performance of a system. This fact has
been widely reported in the NLP literature and also was clearly
observed in the CoNLL-2005 shared task evaluation on Semantic Role

In this direction, we encourage the submission of papers addressing
the portability and adaptation of learning-based systems to changing
application domains. Transfer learning, domain adaptation,
bootstrapping, semi-supervised learning, active learning, etc. are
some keywords that might apply here.

Moreover, the traditional decomposition of natural language processing
into a pipeline of specialized linguistic analyzers can also make
end-to-end systems fragile. The assumption that each level can be
satisfactory resolved before advancing to the following processor is
clearly false given the current state-of-the-art for most
tasks. Experience suggests that error propagation through cascades of
processors may in aggregate severely degrade performance on the final
task. One obvious and appealing solution (but also more complex) is to
try to jointly model several subtasks at the same time, both at the
learning and inference stages. This can allow systems to capture
correlations between stages, searching for global solutions, rather
than greedily maximizing local quality. However, practical constraints
argue that some decomposition is necessary for efficient learning and
inference. Thus, papers addressing the issues involved in processing
across multiple linguistic layers will be also welcome.

Shared Task: Multilingual Dependency Parsing
The shared task of CoNLL-X will be multi-lingual grammatical relation
finding (dependency parsing). Following previous CoNLL shared tasks
(NP bracketing, chunking, clause identification, language independent
named-entity recognition, and semantic role labeling), this task aims
to define and extend the current state of the art in dependency
parsing - a technology which complements the previous tasks by
producing a different kind of syntactic description of input text.

Ideally, a parser should be trainable for any language, possibly by
adjusting a small number of hyperparameters. The CoNLL-X shared task
will provide the community with a benchmark for evaluating their
parsers across different languages. Because of the variety of
languages and the interest in parser performance across languages, the
focus of the CoNLL-X shared task will be on qualitative evaluation
(along with the quantitative scores as before). We will require the
participants to provide an informative error analysis and will
ourselves perform a cross-system comparison. This, we expect, will
result in a clear picture of the problems that lie ahead for
multilingual parsing and the kind of work necessary for adapting
existing parsing architectures across languages.

A detailed description of the shared task and further information
regarding scheduling, datasets, paper submission, etc. are available
from http://www.cnts.ua.ac.be/conll/st.html

Invited Speakers
(to be announced)

Main Session Submissions

A paper submitted to CoNLL-X must describe original, unpublished
work. Submit a full paper of no more than 8 pages in PDF format by
March 5 2006, electronically through the web form at:

Only electronic submissions will be accepted. The submitted paper
should be in two column format and follow the HLT-NAACL style (see
http://nlp.cs.nyu.edu/hlt-naacl06/cfp.html). Authors who cannot submit
a PDF file electronically should contact the program co-chairs.

Since reviewing will be blind, the paper should not include the
authors' names and affiliations, and there should be no
self-references that reveal the authors' identity. In the submission
form, you will be asked for the following information: paper title,
authors' names, affiliations, and email addresses, contact author's
email address, a list of keywords, abstract, and an indication of
whether the paper has been simultaneously submitted to other
conferences (and if so which conferences). The contact author of an
accepted paper under multiple submissions should inform the program
co-chairs immediately whether he or she intends the accepted paper to
appear in CoNLL-X. A paper that appears in CoNLL-X must be withdrawn
from other conferences.

Authors of accepted submissions are to produce a final paper to be
published in the proceedings of the conference, which will be
available at the conference for participants, and distributed
afterwards by ACL. Final papers must follow the HLT-NAACL style and
are due April 21, 2006.

Shared Task Submissions
See the shared task web page (http://www.cnts.ua.ac.be/conll/st.html)
for updated information

Important Dates
Deadline for paper submission: March 5, 2006
Notification of acceptance of papers: April 9, 2006
Deadline for camera-ready papers: April 21, 2006
Conference: June 8-9, 2006

Conference Organizers
Lluís Màrquez
Software Department
Polytechnical University of Catalunya
Barcelona, Catalunya, Spain
lluism (at) lsi.upc.edu

Dan Klein
Computer Science Division
University of California at Berkeley
Berkeley, CA, USA
klein (at) cs.berkeley.edu

Shared Task Organizers
Sabine Buchholz
Toshiba Research Europe Ltd (UK)
sabine.buchholz (at) crl.toshiba.co.uk

Amit Dubey
University of Edinburgh (UK)
adubey (at) inf.ed.ac.uk

Yuval Krymolowski
University of Haifa (Israel)
yuval (at) cs.haifa.ac.il

Erwin Marsi
Tilburg University (The Netherlands)
E.C.Marsi (at) uvt.nl

Information Officer
Erik Tjong Kim Sang
University of Amsterdam (The Netherlands)
erikt (at) science.uva.nl

Program Committee

* Eneko Agirre, University of the Basque Country, Spain
* Regina Barzilay, Massachusetts Institute of Technology, USA
* Thorsten Brants, Google Inc, USA
* Xavier Carreras, Polytechnical University of Catalunya, Spain
* Eugene Charniak, Brown University, USA
* James Cussens, University of York, UK
* Walter Daelemans, University of Antwerp, Belgium
* Radu Florian, IBM, USA
* Dayne Freitag, Fair Isaac Corporation, USA
* Philipp Koehn, University of Edinburgh, UK
* Rob Malouf, San Diego State University, USA
* Yuji Matsumoto, Nara Institute of Science and Technology, Japan
* Andrew McCallum, University of Massachusetts Amherst, USA
* Rada Mihalcea, University of North Texas, USA
* Alessandro Moschitti, University of Rome Tor Vergata, Italy
* John Nerbonne, University of Groningen, The Netherlands
* Hwee-Tou Ng, National University of Singapore, Singapore
* Franz Josef Och, Google, Inc., USA
* Miles Osborne, University of Edinburgh, UK
* David Powers, Flinders University, Australia
* Ellen Riloff, University of Utah, USA
* Dan Roth, University of Illinois at Urbana-Champaign, USA
* Anoop Sarkar, Simon Fraser University, Canada
* Suzanne Stevenson, University of Toronto, Canada
* Mihai Surdeanu, Polytechnical University of Catalunya, Spain
* Charles Sutton, University of Massachusetts Amherst, USA
* Antal van den Bosch, Tilburg University, The Netherlands
* Janyce Wiebe, University of Pittsburgh, USA
* Dekai Wu, The Hong Kong University of Science & Technology, Hong Kong

More information about the Corpora-archive mailing list