<p>-----Apologies for duplicat multiple postings-----</p>
<div>***THIRD CALL FOR PAPERS***</div>
<div><br>Second Workshop on Applying Machine Learning Techniques to Optimise<br>the Division of Labour in Hybrid MT (ML4HMT-12 WS and Shared Task) at<br>COLING 2012</div>
<div> </div>
<div>Mumbai (India), 9th December, 2012 <br>URL: <a href="http://www.dfki.de/ml4hmt/">http://www.dfki.de/ml4hmt/</a></div>
<p>The workshop and associated shared task are an effort to trigger a<br>systematic investigation on improving state-of-the-art hybrid machine<br>translation, making use of advanced machine-learning (ML)<br>methodologies. It follows the ML4HMT-11 workshop which took place last<br>
November in Barcelona. The first workshop also road-tested a shared<br>task (and associated data set) and laid the basis for a broader reach<br>in 2012.</p>
<div>Regular Papers ML4HMT-12</div>
<div><br>We are soliciting original papers on hybrid MT, including (but not<br>limited to):</div>
<div>* use of machine learning methods in hybrid MT;<br>* system combination: parallel in multi-engine MT (MEMT) or sequential <br> in statistical post-editing (SPMT);<br>* combining phrases and translation units from different types of MT;<br>
* syntactic pre-/re-ordering;<br>* using richer linguistic information in phrase-based or in hierarchical <br> SMT;<br>* learning resources (e.g., transfer rules, transduction grammars) for <br> probabilistic rule-based MT.</div>
<div> </div>
<div>Full papers should be anonymous and follow the COLING full paper<br>format (<a href="http://www.coling2012-iitb.org/call_for_papers.php">http://www.coling2012-iitb.org/call_for_papers.php</a>). To submit<br>contributions, please follow the instructions at the Workshop<br>
management system submission website:<br><a href="https://www.softconf.com/coling2012/ML4HMT12/">https://www.softconf.com/coling2012/ML4HMT12/</a>. The contributions will<br>undergo a double-blind review by members of the programme committee.</div>
<p><br>Shared Task ML4HMT-12</p>
<p>The main focus of the Shared Task is to address the question:</p>
<p>-Can Hybrid MT and System Combination techniques benefit from extra<br> information (linguistically motivated, decoding, runtime, confidence<br> scores, or other meta-data) from the systems involved?</p>
<p>Participants are invited to build hybrid MT systems and/or system<br>combinations by using the output of several MT systems of different<br>types, as provided by the organisers. While participants are<br>encouraged to use machine learning techniques to explore the<br>
additional meta-data information sources, other general improvements<br>in hybrid and combination based MT are welcome to participate in the<br>challenge. For systems that exploit additional meta-data information<br>the challenge is that additional meta-data is highly heterogeneous and<br>
(individual) system specific.</p>
<p><br>Data: The ML4HMT-12 Shared Task involves (ES-EN) and (ZH-EN) data<br>sets, in each case translating into EN.</p>
<p><br>* (ES-EN): Participants are given a bilingual tuning set aligned<br> at a sentence level. Each "bilingual sentence" contains: 1) the<br> source sentence, 2) the target (reference) sentence and 3) the<br>
corresponding multiple output translations from four systems, based<br> on different MT approaches (Apertium, Ramirez-Sanchez, 2006; Lucy,<br> Alonso and Thurmair, 2003; Moses, Koehn et. al., 2007). The output<br> has been annotated with system-internal meta-data information<br>
derived from the translation process of each of the systems.</p>
<p>* (ZH-EN) A corresponding data set for ZH-EN with output translations<br> from three systems (Moses, Koehn et. al., 2007;ICT_Chiero, Mi<br> et. al., 2009; and Huajian RBMT) will be provided. (Participants<br> are required to fill out a shared task evaluation agreement form<br>
and obtain the ZH-EN data from LDC).</p>
<div>Participants are challenged to build an MT mechanism where possible<br>making effective use of the system-specific MT meta-data output. They<br>can provide solutions based on opensource systems, or develop their<br>
own mechanisms. The tuning set can be used for tuning the systems or<br>for training the systems. Final submissions have to include<br>translation output on a test set, which will be made available one<br>week after training data release. Data will be provided to build<br>
language/reordering models, possibly re-using existing resources from<br>MT research.</div>
<div><br>Participants can also make use of additional (linguistic analysis,<br>confidence estimation etc.) tools, if their systems require so, but<br>they have to explicitly declare this upon submission, so that they are<br>
judged as "unconstrained" systems. This will allow for a better<br>comparison between participating systems.</div>
<p>System output will be judged via peer-based human evaluation as well<br>as automatic evaluation. During the evaluation phase, participants<br>will be requested to rank system outputs of other participants through<br>a web-based interface (Appraise, Federmann 2010). Automatic metrics<br>
include BLEU (Papineni et. Al, 2002), TER (Snover et al., 2006) and<br>METEOR (Lavie, 2005).</p>
<div> </div>
<div>Shared task participants will be invited to submit system description<br>papers (7 pages, not blind and should follow COLING format,<br><a href="http://www.coling2012-iitb.org/call_for_papers.php">http://www.coling2012-iitb.org/call_for_papers.php</a>).</div>
<p>For submissions, please follow the instructions at the Workshop<br>management system submission<br>website:<a href="https://www.softconf.com/coling2012/ML4HMT12/">https://www.softconf.com/coling2012/ML4HMT12/</a></p>
<p><br>Important Dates 2012</p>
<p><br>15th August: Shared task Tuning data release (updated ML4HMT corpus)<br>23rd August: Shared task Test data release<br>15th September: Shared task Translation results submission deadline<br>21st September: Shared task Evaluation results release<br>
30th September: Workshop full paper and Shared task system description <br>paper submission deadline<br>31st October: Workshop paper accept/reject notification<br>15th November: Workshop and Shared task Camera ready paper due<br>
9th December: ML4HMT-12 Workshop</p>
<p><br>Organizers</p>
<p><br>-Prof. Josef van Genabith, Dublin City University (DCU) and Centre for <br> Next Generation Localisation (CNGL)<br>-Prof. Toni Badia, Universitat Pompeu Fabra and Barcelona Media (BM)<br>-Christian Federmann, German Research Center for Artificial Intelligence <br>
(DFKI), contact <a href="mailto:person%3Acfedermann@dfki.de">person:cfedermann@dfki.de</a><br>-Dr. Maite Melero, Barcelona Media (BM)<br>-Dr. Marta R. Costa-jussa, Barcelona Media (BM)<br>-Dr. Tsuyoshi Okita, Dublin City University (DCU)</p>
<p><br>Program committee</p>
<p><br>- Eleftherios Avramidis (German Research Center for Artificial Intelligence, Germany)<br>- Prof. Sivaji Bandyopadhyay (Jadavpur University, India)<br>- Dr. Rafael Banchs (Institute for Infocomm Research - I2R, Singapore)<br>
- Prof. Loic Barrault (LIUM - University of Le Mans, France)<br>- Prof. Antal van den Bosch (Centre for Language Studies, Radboud University Nijmegen, Netherlands)<br>- Dr. Grzegorz Chrupala (Saarland University, Saarbrucken, Germany)<br>
- Prof. Jinhua Du (Xi'an University of Technology (XAUT), China)<br>- Dr. Andreas Eisele (Directorate-General for Translation (DGT), Luxembourg)<br>- Dr. Cristina Espana-Bonet (Technical University of Catalonia, TALP, Barcelona)<br>
- Dr. Declan Groves (Center for Next Generation Localisation, Dublin City University, Ireland)<br>- Prof. Jan Hajic (Institute of Formal and Applied Linguistics, Charles University in Prague)<br>- Prof. Timo Honkela (Aalto University, Finland)<br>
- Dr. Patrick Lambert (LIUM - University of Le Mans, France)<br>- Prof. Qun Liu (Institute of Computing Technology, Chinese Academy of Sciences, China)<br>- Dr. Maite Melero (Barcelona Media Innovation Center, Spain)<br>- Dr. Tsuyoshi Okita (Dublin City University, Ireland)<br>
- Prof. Pavel Pecina (Institute of Formal and Applied Linguistics, Charles University in Prague)<br>- Dr. Marta R. Costa-jussa (Barcelona Media Innovation Center, Spain)<br>- Dr. Felipe Sanchez Martinez (Escuela Politecnica Superior, Universidad de Alicante, Spain)<br>
- Dr. Nicolas Stroppa (Google, Zurich, Switzerland)<br>- Prof. Hans Uszkoreit (German Research Center for Artificial Intelligence, Germany)<br>- Dr. David Vilar (German Research Center for Artificial Intelligence, Germany)</p>
<p><br>The ML4HMT workshop is supported by the META-NET T4ME project<br>(<a href="http://www.meta-net.eu/">http://www.meta-net.eu/</a>), funded by the DG INFSO of the European<br>Commission through the Seventh Framework Programme, grant agreement<br>
no.: 249119.<br></p>