<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META content="text/html; charset=iso-8859-1" http-equiv=Content-Type>
<META name=GENERATOR content="MSHTML 8.00.6001.19170">
<STYLE></STYLE>
</HEAD>
<BODY bgColor=#ffffff>Apologies for multiple postings<BR>Please distribute to
colleagues<BR><BR>============================================================<BR><BR>
5th WORKSHOP ON BUILDING AND USING COMPARABLE CORPORA<BR><BR>
Language Resources for Machine Translation<BR> in
Less-Resourced Languages and Domains<BR><BR> Co-located with
LREC 2012<BR> Lütfi Kirdar Istanbul Exhibition and Congress
Centre<BR> Saturday, 26 May 2012<BR><BR>
DEADLINE FOR PAPERS: 15 February 2012<BR><BR> <A
href="http://hnk.ffzg.hr/5bucc2012">http://hnk.ffzg.hr/5bucc2012</A><BR><BR>
Endorsed by<BR> * ACL SIGWAC (Special Interest Group on
Web as Corpus)<BR> * FLaReNet (Fostering Language
Resources
Network)<BR><BR>============================================================<BR><BR>MOTIVATION<BR><BR>In
the language engineering and the linguistics communities,<BR>research in
comparable corpora has been motivated by two main<BR>reasons. In language
engineering, it is chiefly motivated by the<BR>need to use comparable corpora as
training data for statistical<BR>NLP applications such as statistical machine
translation or<BR>cross-lingual retrieval. In linguistics, on the other
hand,<BR>comparable corpora are of interest in themselves by making<BR>possible
inter-linguistic discoveries and comparisons. It is<BR>generally accepted in
both communities that comparable corpora<BR>are documents in one or several
languages that are comparable in<BR>content and form in various degrees and
dimensions. We believe<BR>that the linguistic definitions and observations
related to<BR>comparable corpora can improve methods to mine such corpora
for<BR>applications of statistical NLP. As such, it is of great interest<BR>to
bring together builders and users of such corpora.<BR><BR>The scarcity of
parallel corpora has motivated research concerning<BR>the use of comparable
corpora: pairs of monolingual corpora selected<BR>according to the same set of
criteria, but in different languages<BR>or language varieties. Non-parallel yet
comparable corpora overcome<BR>the two limitations of parallel corpora, since
sources for original,<BR>monolingual texts are much more abundant than
translated texts.<BR>However, because of their nature, mining translations in
comparable<BR>corpora is much more challenging than in parallel corpora.
What<BR>constitutes a good comparable corpus, for a given task or per
se,<BR>also requires specific attention: while the definition of a
parallel<BR>corpus is fairly straightforward, building a non-parallel
corpus<BR>requires control over the selection of source texts in both
languages.<BR><BR>Parallel corpora are a key resource as training data for
statistical<BR>machine translation, and for building or extending bilingual
lexicons<BR>and terminologies. However, beyond a few language pairs such as
English-<BR>French or English-Chinese and a few contexts such as parliamentary
debates<BR>or legal texts, they remain a scarce resource, despite the creation
of<BR>automated methods to collect parallel corpora from the Web. To
exemplify<BR>such issues in a practical setting, this year's special focus will
be on<BR><BR> Language Resources for Machine
Translation<BR> in Less-Resourced Languages and Domains<BR><BR>with
the aim of overcoming the shortage of parallel resources<BR>when building MT
systems for less-resourced languages and domains,<BR>particularly by usage of
comparable corpora for finding parallel data<BR>within and by reaching out for
"hidden" parallel data. Lack of sufficient<BR>language resources for many
language pairs and domains is currently one<BR>of the major obstacles in further
advancement of machine translation.<BR><BR><BR>TOPICS<BR><BR>We solicit
contributions including but not limited to the following topics:<BR><BR>Topics
related to the special theme:<BR><BR>* comparable corpora use in MT<BR>*
comparable corpora processing tools/kits for MT<BR>* parallel corpora usage<BR>*
parallel corpora processing tools/platforms<BR>* MT for less-resourced
languages<BR>* MT for less-resourced domains<BR>* open source SMT systems
(Moses, etc.)<BR>* publicly available SMT<BR><BR>Building Comparable
Corpora:<BR><BR> * Human translations<BR> * Automatic and
semi-automatic methods<BR> * Methods to mine parallel and non-parallel
corpora from the Web<BR> * Tools and criteria to evaluate the comparability
of corpora<BR> * Parallel vs non-parallel corpora, monolingual
corpora<BR> * Rare and minority languages<BR> * Across language
families<BR> * Multi-media/multi-modal comparable
corpora<BR><BR>Applications of comparable corpora:<BR><BR> * Human
translations<BR> * Language learning<BR> * Cross-language information
retrieval & document categorization<BR> * Bilingual
projections<BR> * Machine translation<BR> * Writing
assistance<BR><BR>Mining from Comparable Corpora:<BR><BR> * Extraction of
parallel segments or paraphrases from comparable<BR>
corpora<BR> * Extraction of bilingual and multilingual translations of
single<BR> words and multi-word expressions; proper names, named
entities,<BR> etc.<BR><BR><BR>IMPORTANT DATES
(TENTATIVE)<BR><BR> 15 February 2012 Deadline for
submission of full papers<BR> 10 March
2012 Notification of acceptance<BR> 20
March 2012 Camera-ready papers
due<BR> 26 May 2012
Workshop date<BR><BR><BR>SUBMISSION INFORMATION<BR><BR>Papers should follow the
LREC main conference formatting details (to be<BR>announced on the conference
website <A
href="http://www.lrec-conf.org/lrec2012/">http://www.lrec-conf.org/lrec2012/</A>)<BR>and
should be submitted as a PDF-file of no more than ten pages via the<BR>START
workshop manager: <A
href="https://www.softconf.com/lrec2012/BUCC2012/">https://www.softconf.com/lrec2012/BUCC2012/</A><BR>Reviewing
will be double blind, so the papers should not reveal the<BR>authors' identity.
Accepted papers will be published in the workshop<BR>proceedings.<BR><BR>Double
submission policy: Parallel submission to other meetings or<BR>publications are
possible but must be immediately notified to the<BR>workshop
organizers.<BR><BR>When submitting a paper through the START page, authors will
be asked<BR>to provide information about the resources that have been used for
the work<BR>described in their paper or are an outcome of their research. For
details on<BR>this initiative, please refer to <A
href="http://www.lrec-conf.org/lrec2012/?LRE-Map-2012">http://www.lrec-conf.org/lrec2012/?LRE-Map-2012</A>.<BR>Authors
will also be asked to contribute to the Language Library, the new<BR>initiative
of LREC 2012.<BR><BR>For further information, please contact<BR>
Reinhard Rapp reinhardrapp (at) gmx (dot) de<BR> or Marko Tadic
marko.tadic (at) ffzg (dot) hr<BR><BR><BR>ORGANISERS<BR><BR> Reinhard
Rapp, Universities of Mainz (Germany)and Leeds (UK)<BR> Marko Tadic,
University of Zagreb (Croatia)<BR> Serge Sharoff, University of Leeds
(UK)<BR> Andrejs Vasiljevs, Tilde SIA, Riga, Latvia<BR> Pierre
Zweigenbaum, LIMSI, CNRS, Orsay, and ERTIM, INALCO, Paris
(France)<BR><BR><BR>SCIENTIFIC COMMITTEE<BR><BR>* Srinivas Bangalore (AT&T
Labs, USA)<BR>* Caroline Barrière (National Research Council Canada)<BR>* Chris
Biemann (Microsoft / Powerset, San Francisco, USA)<BR>* Lynne Bowker (University
of Ottawa, Canada)<BR>* Hervé Déjean (Xerox Research Centre Europe, Grenoble,
France)<BR>* Andreas Eisele (DFKI, Saarbrücken, Germany)<BR>* Rob Gaizauskas
(University of Sheffield, UK)<BR>* Éric Gaussier (Université Joseph Fourier,
Grenoble, France)<BR>* Nikos Glaros (ILSP, Athens, Greece)<BR>* Gregory
Grefenstette (Exalead/Dassault Systemes, Paris, France)<BR>* Silvia
Hansen-Schirra (University of Mainz, Germany)<BR>* Kyo Kageura (University of
Tokyo, Japan)<BR>* Adam Kilgarriff (Lexical Computing Ltd, UK)<BR>* Natalie
Kübler (Université Paris Diderot, France)<BR>* Philippe Langlais (Université de
Montréal, Canada)<BR>* Tony McEnery (Lancaster University, UK)<BR>* Emmanuel
Morin (Université de Nantes, France)<BR>* Dragos Stefan Munteanu (Language
Weaver Inc., USA)<BR>* Lene Offersgaard (University of Copenhagen, Denmark)<BR>*
Reinhard Rapp (Universities of Mainz, Germany, and Leeds, UK)<BR>* Sujith Ravi
(Yahoo! Research, Santa Clara, CA, USA)<BR>* Serge Sharoff (University of Leeds,
UK)<BR>* Michel Simard (National Research Council Canada)<BR>* Inguna Skadina
(Tilde, Riga, Latvia)<BR>* Monique Slodzian (INALCO, Paris, France)<BR>*
Benjamin Tsou (The Hong Kong Institute of Education, China)<BR>* Dan Tufis
(Romanian Academy, Bucharest, Romania)<BR>* Justin Washtell (University of
Leeds, UK)<BR>* Oliver Wilson (University of Edinburgh, UK)<BR>* Michael Zock
(LIF, CNRS Marseille, France)<BR>* Pierre Zweigenbaum (LIMSI-CNRS, Orsay,
France)<BR></BODY></HTML>