Machine Translation @ the National Centre for Language Technology

Dublin City University, Ireland

Home People Projects Publications Events Announcements Links
-- current projects --
  EuroMatrix+
  Panacea
  CoSyne
  PLuTO
  T4ME
  Confident MT

-- completed projects --
  Prospect
  Attempt
  Sign language
    translation

  Evaluation
  Transbooster
  DVD subtitling
  LFG-DOP
  EBMT & Marker
    Hypothesis

  DOP & DOT
  Hybrid EBMT-SMT
Title:Translating DVD Subtitles using Example-Based Machine Translation
Duration:October 1st 2005 - September 30th 2006
Funded by: Enterprise Ireland's Commercialisation Funding Scheme
People:Stephen Armstrong, Colm Caffrey, Marian Flanagan, Dorothy Kenny, Minako O'Hagan, Andy Way
Description:Due to limited budgets and an ever-diminishing time-frame for the production of subtitles for films released in cinemas and on DVDs, there is a compelling case for a technology-based translation solution for subtitles. The aim of this project is to develop an Example-Based Machine Translation (EBMT) system for the translation of English DVD subtitles into German and Japanese.
 To our knowledge this is the first time that any MT approach has been used with DVD subtitle translation. Due to the specific language rules imposed on the subtitler, subtitles are particularly well suited to EBMT. In addition to this, recent studies in EBMT have shown how it outperformed online rule-based MT systems as well as statistical models of rule-based Machine Translation.
 For our corpus we have acquired sets of both homogeneous and heterogeneous data from a wealth of sources including existing subtitle files, eBooks, a Translation Memory, and Speech-to-text transcripts. During testing and evaluation of our system we intend to 'plug in and out' the different sets of data, and analyse the output translation of each set. In addition to MT evaluation metrics such as BLEU and NIST, we will conduct a real-user evaluation, as this is the only reliable way of testing whether our system is working at a level, which would be suitable to produce subtitles for commercial markets, and accepted by the target audience.
 
Related Sites: NCLT | School of Computing | School of Applied Languages and Intercultural Studies | Dublin City University
Last update: June 16 2007