CFP: Machine Translation Journal

15 Sep 2012
Europe/Stockholm

Special Issue on Quality Estimation http://www.springer.com/computer/artificial/journal/10590

Guest editors: Lucia Specia (University of Sheffield) Radu Soricut (SDL Language Weaver)


Quality estimation is a topic of increasing interest in the field of Machine Translation (MT). It aims at providing a quality indicator for unseen translated texts at various granularity levels. It has the potential to make MT more useful in a number of practical scenarios, including:

  • Deciding whether a given translation is good enough for publishing as is
  • Informing readers whether or not they can rely on a translation
  • Filtering out translations that are not good enough for post-editing by professional translators
  • Selecting the best translation among options from multiple translation systems

Quality estimation techniques can also be useful for estimating the usefulness of human translations for specific purposes, including the selection of relevant/domain-specific parallel data to build MT systems.

A shared-task on quality estimation has been organised as part of WMT-12 (http://www.statmt.org/wmt12/quality-estimation-task.html) to put together researchers in the field and provide a first common ground for development and comparison of quality estimation systems. It included the release of training and test sets, along with novel evaluation metrics and a baseline system. This shared task attracted participants from eleven teams around the world, and resulted in a number of systems proposing innovative ways of addressing this challenging problem.

This special issue invites papers describing novel approaches for quality estimation as well as interesting problems and challenges in the field. Extended versions of systems presented at the shared task are welcome, but submissions are open to everyone (not only to task participants). We particularly encourage submissions on:

  • Approaches for word-/phrase-/sentence-/document-level quality estimation
  • Studies on boundaries between quality estimation and error detection
  • Interesting applications of quality estimation
  • Evaluation strategies/practices/metrics for quality estimation
  • Different learning paradigms for the problem (ranking vs scoring vs binary decisions, etc)
  • Studies on correlating human judgements and automatic measurements of MT quality in the context of quality estimation
  • Domain adaptation for quality estimation

Contributions from approaches which are not yet able to outperform the state of the art systems are also suitable for inclusion in this issue, provided they stem from sound directions with clear potential for further developments.


Submission guidelines:


Important dates:

  • Paper submission: September 15, 2012
  • Notification to authors: October 30, 2012
  • Camera-ready*: December 31, 2012

  • tentative - depending on the number of review rounds necessary