Account Options

  1. Sign in
The old Google Groups will be going away soon.
Switch to the new Google Groups.
Google Groups Home
« Groups Home
judging edited machine translation output
There are currently too many topics in this group that display first. To make this topic appear first, remove this option from another topic.
There was an error processing your request. Please try again.
flag
  4 messages - Collapse all  -  Translate all to Translated (View all originals)
The group you are posting to is a Usenet group. Messages posted to this group will make your email address visible to anyone on the Internet.
Your reply message has not been sent.
Your post will appear after it is approved by moderators
 
From:
To:
Cc:
Followup To:
Add Cc | Add Followup-to | Edit Subject
Subject:
Validation:
For verification purposes please type the characters you see in the picture below or the numbers you hear by clicking the accessibility icon. Listen and type the numbers you hear
 
Chris Callison-Burch  
View profile  
 More options Jan 27 2009, 11:21 am
From: Chris Callison-Burch <c...@cs.jhu.edu>
Date: Tue, 27 Jan 2009 17:21:12 +0100
Local: Tues, Jan 27 2009 11:21 am
Subject: judging edited machine translation output
We have updated the WMT09 manual evaluation tasks.  Now, instead of  
editing the output of machine translation systems, we are asking you  
to judge whether the edited translations are correct or not.

The new task shows the source and the reference along with some  
context.  You should supply simple Yes/No judgments about whether the  
edited translations represent fully fluent and meaning-equivalent  
alternatives to the reference.

The deadline for the manual evaluation is the end of Friday.  If you  
participated in the shared task, and your group has not yet completed  
your 8 hour time commitment, then please log in and judge some  
translations.  The acceptability judgments are highest priority, but  
the ranking statistics are still useful.
Here is the URL for the manual evaluation:
         http://statmt.org/wmt09/judge/

Thanks again for your participation!

Chris Callison-Burch
Philipp Koehn
Christof Monz
Josh Schroeder


 
You must Sign in before you can post messages.
To post a message you must first join this group.
Please update your nickname on the subscription settings page before posting.
You do not have the permission required to post.
Chris Callison-Burch  
View profile  
 More options Jan 28 2009, 3:19 am
From: Chris Callison-Burch <c...@cs.jhu.edu>
Date: Wed, 28 Jan 2009 09:19:52 +0100
Local: Wed, Jan 28 2009 3:19 am
Subject: Re: judging edited machine translation output
Hi Gregor,

Good point: if you know the source language, feel free to use that  
instead of, or in addition to, the reference.

--Chris

On Jan 27, 2009, at 6:40 PM, in...@filmstudio.rwth-aachen.de wrote:


 
You must Sign in before you can post messages.
To post a message you must first join this group.
Please update your nickname on the subscription settings page before posting.
You do not have the permission required to post.
Yuval Marton  
View profile  
 More options Jan 28 2009, 12:47 pm
From: Yuval Marton <yuvalmar...@gmail.com>
Date: Wed, 28 Jan 2009 12:47:57 -0500
Local: Wed, Jan 28 2009 12:47 pm
Subject: Re: judging edited machine translation output

Hi Chris,
Following the previous concern about the references, I have to say that
reading some of the references makes me wonder if they were produced by
humans (I hope I am not offending anyone here! :-)   Some references, I
think, were not very fluent, and/or not very faithful to the source. For
example :
In the production of metal processing a price increase of 13.2 percent, in
the field of food, beverage, tobacco production 11 percent growth was seen.
(source: En el sector de la elaboración del metal se observó un aumento de
un 13,2 por ciento, mientras que en la industria de alimentos, bebidas y
tabacos, de peso determinante, hubo un aumento de 11 por ciento. )

or:
"I want to be a young mother, doesn't matter the candidate"
(source: "Quiero ser mamá joven; el candidato no me preocupa" )

So I decided to use my non-native (but hopefully good enough) English, and
very poor Spanish, in order to judge translations as "Yes" if they were
around the same quality as the reference (or better), and not vote them down
if I thought the reference was not better.  Please feel free to ignore my
judgments if you disagree with this policy.

-Yuval

On Wed, Jan 28, 2009 at 3:19 AM, Chris Callison-Burch <c...@cs.jhu.edu>wrote:


 
You must Sign in before you can post messages.
To post a message you must first join this group.
Please update your nickname on the subscription settings page before posting.
You do not have the permission required to post.
Philipp Koehn  
View profile  
 More options Jan 28 2009, 4:00 pm
From: Philipp Koehn <pko...@inf.ed.ac.uk>
Date: Wed, 28 Jan 2009 21:00:57 +0000
Local: Wed, Jan 28 2009 4:00 pm
Subject: Re: judging edited machine translation output
Hi,

if you can read the source, you should ignore the reference - but
you should not judge a faulty machine translation as good just because
the human translator did not do better.

-phi


 
You must Sign in before you can post messages.
To post a message you must first join this group.
Please update your nickname on the subscription settings page before posting.
You do not have the permission required to post.
End of messages
« Back to Discussions « Newer topic     Older topic »