search

cifkao / tct

Twitter Crowd Translation – infrastructure for human and machine tweet translation
GNU General Public License v2.0
1 stars 0 forks source link

small dictionary into e-mails requesting translations #33

Open obo opened 9 years ago

obo commented 9 years ago

It would be very useful (and a further great incentive for learners of source languages to join us as translators) to have a small dictionary with just the words from the sentence in the e-mail where we are requesting translations.

If we stretch the idea far, we should also provide links to relevant knowledge sources so that people can check and disambiguate.

For example:

Please translate into Czech: I stressed to UN Secretary General the need to start the political process of releasing hostages,including Nadiya Savchenko and Oleh Sentsov

I ?= já, ... stressed ?= ... UN ?= United Nations, ... [abbreviation expanded, e.g. from http://www.abbreviations.com/UN] Secretary ?= ... UN Secretary General ?= Generální tajemník OSN (http://en.wikipedia.org/wiki/Secretary-General_of_the_United_Nations) ... hostages = rukojmí, ... Nadiya Savchenko ?= Nadija Savčenková (http://en.wikipedia.org/wiki/Nadiya_Savchenko) ...

My example combines single-word items (which we get from lex.f2e and e2f Moses tables for free) with random Internet resources. We could also use highly probably multi-word phrases from Moses translation options...

edasubert commented 9 years ago

Notes from talk on post editing (not everything relevant to tct): PE is faster and of better quality PE is more cognitively demanding and hated by translators

Keywords: fear, pride, satisfaction, money, stress, knowledge

Recommendation:

  1. Improve machine translation :-)
  2. Develop accurate and meaningful quality metrics
  3. improve system ability to learn (limit repetitivness)
  4. Include translators in research

Probably won't be suitable for experienced translators but might be great guide for beginners