HaCaT 2014 : Human in the Loop: 1st Workshop on Humans and Computer-asisted Translation
Call For Papers
Collocated with EACL 2014
The past decade has seen tremendous progress in the speed, quality and availability of automatic translation of natural languages. While automatic translation quality still regularly falls short of publication or near-publication quality, contemporary machine translation can deliver a level of quality that may boost the productivity of human translators by providing them with raw translations to work from, ensuring consistency in terminology, and fast access to terminological databases and databases of previous translations in the form of translation memories and bilingual concordances.
Much research in the CL community in recent years has focused on improving fully automatic MT, but we still know comparatively little about how humans translate, and how to optimally organize human-machine interaction in computer-assisted translation. This workshop aims to provide a platform to discuss these issues and to present empirical results, data sets, case studies, and tools for computer-assisted translation and the study of cognitive processes in both fully human and computer-assisted human translation. We particularly encourage submissions related to the following
Topics of interest
Topics of interest include, but are not limited to, addressing questions such as:
* Behavioral studies of human translators in action. What do they spend their time on? Where do they get stuck? How much context do they need? At what level of understanding do they process the text?
* CAT tools and tools for the study of human translation.
* Features and functions of a CAT user interface. What features do translators require a UI to have? What do they need for a CAT UI to be productive?
* Data sets, especially data sets showing how translations evolve during multi-stage processing in a professional setting (e.g., from junior translator to senior translator to final revision).
* In-depth analyses of both human and automatic translations. What are typical errors? Do professional translators work very differently to casual translators?
* Novel types of assistance for human translators.
* Adapting machine translation systems to the needs of human translators.
* The CAT back-end: SMT vs translation memories and example-based MT, hierarchical vs. phrase-based SMT, other approaches and combinations thereof. What do translators prefer? What are they most productive with?
* Revising the human-computer interaction scheme. How can the machine translation system learn from the user's feedback? Are other interaction protocols such as interactive machine translation better suited? How can an e-pen, a gaze tracking device, a speech recognition system or other multi-modal devices improve translation productivity?
We invite three types of submissions. All must present original, previously unpublished work. Double submission of long or short papers to other conferences and workshops is acceptable as long as this disclosed at the time of submission. It is expected that the workshop organizers are notified immediately if the paper will be published elsewhere. Double submission of demos is allowed.
* long papers that present substantial, completed pieces of work. Long papers can consist of up to 8 pages of text (including tables and figures) and up to two 2 extra pages for references.
* short papers that report on work in progress or pilot studies and small studies. Short papers can consist of up to 5 pages of of text (including tables and figures) plus 1 extra page for references.
* system demos that present CAT systems and experimental setups that showcase novel approaches and new features in human-computer interaction in the context of automated and computer-assisted translation. System demos should be submitted in the same format as the short papers (i.e., maximum 5 pages of text plus 1 extra page for references).
Submissions will be handled electronically via the START Conference Manager System. A link to the specific URL will be posted shortly on the workshop web site. Submissions must be anonymized for double-blind reviewing (i.e., not reveal the identity of the author in any form, including meta-data stored in the file; see the file eacl2014.pdf, which is part of the package linked to below) and be in portable document format (PDF). They should follow the general EACL style guidelines available at http://www.eacl2014.org/files/eacl-2014-styles.zip.
Each submission will be blind reviewed at least by two members of the program committee. Accepted papers will be published in the workshop proceedings. In addition, we aim to provide opportunity to publish extended versions of exceptional papers in a special journal edition.
23 January 2014: Paper submission deadline
20 February 2014: Notification of acceptance
3 March 2014: Camera-ready papers due
26 April 2014: EACL Workshop in Gothenburg, Sweden, co-located with EACL 2014.
* Ulrich Germann, University of Edinburgh
* Michael Carl, Copenhagen Business School
* Philipp Koehn, University of Edinburgh
* Germán Sanchis-Trilles, Universitat Politècnica de València
* Francisco Casacuberta, Universitat Politècnica de València
* Robin Hill, University of Edinburgh
* Sharon O'Brien, Dublin City University
Organizing committee, plus
* Bartolomé Mesa-Lao
* Marco Turchi
* Sara Stymne
* Roland Kuhn
* Pascual Martinez
* Barry Haddow
* Fred Hollowood
* Marcello Federico
* Mauro Cettolo
* Matteo Negri
* Pierrette Bouillon
* Manny Rayner
* Philippe Langlais
* Guy Lapalme
* George Foster
* Johanna Gerlach
* Michel Simard
* Marine Carpuat
* Violeta Seretan
* Robert Frederking
* John Moran
* Srinivas Bangalore
* Fabio Alves
* Daniel Ortiz-Martínez
* Vicent Alabau
* Luis A. Leiva
* Jesús González-Rubio
* Nicola Bertoldi
* Christian Buck
* Alessandro Cattelan
* Nicola Bertoldi
* Christophe Servan