5 ms·

This type of crowdsourcing meets ML is a really nice example of where we can leverage humans and machines to the best of their current abilities. It would be g

by etrain 13y ago

This type of crowdsourcing meets ML is a really nice example of where we can leverage humans and machines to the best of their current abilities. It would be great if the feedback from the human workers could get reintegrated into the translation models in an online fashion so that they get better over time. I realize they're probably outsourcing their machine translation, but that would be a terrific fully integrated pipeline.

gracaninja 13y ago
I am Unbabel's CTO. Thanks for your comment. Our goal is exactly to use that feedback to improve our MT systems. We are currently outsourcing our MT, while training our own models (using Moses). Besides generating parallel text, the types of data we will be collecting (e.g. chain of editions performed by each editor on a task), will allow new and interesting algorithms to update the translation models.
- sushirain 13y ago
  I remember reading your NLP papers back in my academic days. Great work. One typo: in the https://www.unbabel.com/pricing/ https://www.unbabel.com/pricing/ page "from Portuguese to Portuguese can take some time".
  - Noxchi 13y ago
    And Italian to Portuguese should be not available.
    - vasco_ 13y ago
      Why should it not be available?
      - byoung2 13y ago
        There is currently a gray dot, which for the other languages is used to indicate that the source and destination language is the same. For Italian to Portuguese this is not the case. It should have one of the other colored icons.
        vasco_ 13y ago
        Ah, thanks, sorry, bug on our part, being corrected right now, thank you for pointing it out.
- moolriaz 13y ago
  Awesome. I'm learning about Moses at the moment (MT course at Edinburgh), see you worked with Ben Taskar also, RIP. Exciting idea here!
- donw 13y ago
  Maybe this is intentional, but from your homepage, I have no idea what languages you handle.
  - vasco_ 13y ago
    Good point, we are in the process of updating our homepage. Things have been moving so fast it is hard to keep up. In any case we currently offer translations in English, Portuguese, Spanish, Italian, French and Turkish. More languages to come soon.
- mtrimpe 13y ago
  Do you offer support for time-coded transcriptions (i.e. video subtitles) as well? We would love a good provider to enrich video with translations (for when we find a good one that offers machine transcription.)
  - gracaninja 13y ago
    Hello, Not yet, but could you send us an example?
    - mtrimpe 13y ago
      We haven't found a good provider yet to do this properly for our use case, but SpeakerText, Koemei and VoiceBase are examples of companies that offer these functionalities. Unfortunately SpeakerText doesn't offer non-post-processed prices, Koemei integrated it into their own product and VoiceBase didn't offer post-processing on request, which we would need for integration into our product. Which format will become mainstream probably depends on HTML5 adoption, which is detailed here http://www.3playmedia.com/how-it-works/how-to-guides/html5-video-captioning/ http://www.3playmedia.com/how-it-works/how-to-guides/html5-v... Currently WebVTT seems to be in the lead. Those formats don't accommodate for timestamps per spoken word though, which would be possible with machine transcription and which I would pay a premium for.
bederoso 13y ago
Duolingo has an also cool approach, where they offer free language courses for students, and charge companies that want certain documents translated. The students themselves translate those documents as part of the course, and once a consensus is reached (a given amount of students came up with the same translation) the document is returned to the submitter, who pays for its translation. http://www.duolingo.com/info http://www.duolingo.com/info Almost like how recaptcha works :-)
- vasco_ 13y ago
  Both recaptcha and duolingo were started by the brilliant Louis Von Ahn. I am a big fan of his work. He was the inspiration for my first startup, while at CMU.

by etrain 13y ago

gracaninja 13y ago

I am Unbabel's CTO. Thanks for your comment. Our goal is exactly to use that feedback to improve our MT systems. We are currently outsourcing our MT, while training our own models (using Moses). Besides generating parallel text, the types of data we will be collecting (e.g. chain of editions performed by each editor on a task), will allow new and interesting algorithms to update the translation models.

sushirain 13y ago
I remember reading your NLP papers back in my academic days. Great work. One typo: in the https://www.unbabel.com/pricing/ https://www.unbabel.com/pricing/ page "from Portuguese to Portuguese can take some time".
- Noxchi 13y ago
  And Italian to Portuguese should be not available.
  - vasco_ 13y ago
    Why should it not be available?
    - byoung2 13y ago
      There is currently a gray dot, which for the other languages is used to indicate that the source and destination language is the same. For Italian to Portuguese this is not the case. It should have one of the other colored icons.
      - vasco_ 13y ago
        Ah, thanks, sorry, bug on our part, being corrected right now, thank you for pointing it out.
moolriaz 13y ago
Awesome. I'm learning about Moses at the moment (MT course at Edinburgh), see you worked with Ben Taskar also, RIP. Exciting idea here!
donw 13y ago
Maybe this is intentional, but from your homepage, I have no idea what languages you handle.
- vasco_ 13y ago
  Good point, we are in the process of updating our homepage. Things have been moving so fast it is hard to keep up. In any case we currently offer translations in English, Portuguese, Spanish, Italian, French and Turkish. More languages to come soon.
mtrimpe 13y ago
Do you offer support for time-coded transcriptions (i.e. video subtitles) as well? We would love a good provider to enrich video with translations (for when we find a good one that offers machine transcription.)
- gracaninja 13y ago
  Hello, Not yet, but could you send us an example?
  - mtrimpe 13y ago
    We haven't found a good provider yet to do this properly for our use case, but SpeakerText, Koemei and VoiceBase are examples of companies that offer these functionalities. Unfortunately SpeakerText doesn't offer non-post-processed prices, Koemei integrated it into their own product and VoiceBase didn't offer post-processing on request, which we would need for integration into our product. Which format will become mainstream probably depends on HTML5 adoption, which is detailed here http://www.3playmedia.com/how-it-works/how-to-guides/html5-video-captioning/ http://www.3playmedia.com/how-it-works/how-to-guides/html5-v... Currently WebVTT seems to be in the lead. Those formats don't accommodate for timestamps per spoken word though, which would be possible with machine transcription and which I would pay a premium for.

bederoso 13y ago

Duolingo has an also cool approach, where they offer free language courses for students, and charge companies that want certain documents translated. The students themselves translate those documents as part of the course, and once a consensus is reached (a given amount of students came up with the same translation) the document is returned to the submitter, who pays for its translation. http://www.duolingo.com/info http://www.duolingo.com/info Almost like how recaptcha works :-)

vasco_ 13y ago
Both recaptcha and duolingo were started by the brilliant Louis Von Ahn. I am a big fan of his work. He was the inspiration for my first startup, while at CMU.