![]() |
| |||||||||||||||
IWSLT 2009 : 6th International Workshop on Spoken Language Translation | |||||||||||||||
Link: http://mastarpj.nict.go.jp/IWSLT2009 | |||||||||||||||
| |||||||||||||||
Call For Papers | |||||||||||||||
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
6th International Workshop on Spoken Language Translation (IWSLT 2009) First Call for Participants / Papers December 1-2, 2009 Tokyo, Japan http://mastarpj.nict.go.jp/IWSLT2009 -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- The 6th International Workshop on Spoken Language Translation (IWSLT) will take place in Tokyo, Japan in December 2009. IWSLT is a yearly, open evaluation campaign for spoken language translation followed by a scientific workshop, in which (1) system descriptions and (2) scientific papers are presented. IWSLT's evaluations are friendly, competition-oriented, with the aim to foster cooperative work and scientific exchange. In this respect, IWSLT proposes challenging research tasks and an open experimental infrastructure including supplied corpora, automatic evaluation servers, and human evaluations for the scientific community working on spoken language translation. The focus of this year's evaluation campaign will be the translation of task-oriented human dialogs in travel situations. The speech data was recorded through human interpreters, where native speakers of different languages were asked to complete certain travel-related tasks like "hotel reservations" using their mother-tongue. The translation of the freely- uttered conversation was carried-out by human interpreters. The obtained speech data was annotated with dialog and speaker information. For the Challenge Task, IWSLT participants will have to translate both, the Chinese and English, outputs of the automatic speech recognizers (lattice, N/1BEST) into English and Chinese, respectively. Like in previous IWSLT events, a standard BTEC Task will be provided. However, the BTEC Task will focus on text input only, i.e. no automatic speech recognition results (lattice, N/1BEST) have to be translated. In addition to the Arabic-English and Chinese-English translation tasks, this year's evaluation campaign features a new input language. Turkish will be used for the first time as an input language for the BTEC translation task. In addition to the evaluation campaign, the IWSLT 2009 workshop also invites scientific paper submissions related to spoken language technologies. Possible topics include, but are not limited to: * Spoken dialog translation * Integration of ASR and MT * SMT, EBMT, RBMT, Hybrid MT * MT evaluation * Language resources for MT * Open source software for MT * Pivot Translation * Task adaptation and portability === Important Dates: *** (1) Evaluation Campaign *** May 22, 2009 -- System Registration Open Jun 19, 2009 -- Training/Develop Corpus Release Aug 14, 2009 -- Test Corpus Release Aug 28, 2009 -- Run Submission Due Sep 11, 2009 -- Result Feedback to Participants Sep 18, 2009 -- Paper Submission Due Oct 16, 2009 -- Notification of Acceptance Oct 31, 2009 -- Camera-ready Submission Due Dec 1-2, 2009 -- Workshop *** (2) Scientific Papers *** Jul 31, 2009 -- Paper Submission Due Sep 25, 2009 -- Notification of Acceptance Oct 31, 2009 -- Camera-ready Submission Due Dec 1-2, 2009 -- Workshop === Organizers Alex Waibel (CMU, USA / UKA, Germany) Marcello Federico (FBK, Italy) Satoshi Nakamura (NICT, Japan) === Chairs Workshop Chair: Eiichiro Sumita (NICT, Japan) Evaluation Campaign Chair: Michael Paul (NICT, Japan) Technical Paper Chair: Marcello Federico (FBK, Italy) === Contact Eiichiro Sumita e-mail: iwsltlocal09@khn.nict.go.jp National Institute of Information and Communications Technology (NICT) 2-2-2 Hikaridai, Keihanna Science City, Kyoto 619-0288 Japan -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- |
|