Re: [gutvol-d] google and the translation thing
Maybe one last word here. The EU use MT technologies to translate the bulk. Yet, the produced texts are still manually processed by humans to get it right. If the google method was so good the EU would not need translators since thier written texts as basically similar in all langauges. They are basically formal debates and legistative in form. Keith.
On Mon, 13 Mar 2006 10:20:14 +0100, "Keith J. Schultz" <schultzk@uni-trier.de> wrote: |Maybe one last word here. | | | The EU use MT technologies to translate the bulk. Yet, the produced | texts are still manually processed by humans to get it right. | If the google method was so good the EU would not need translators | since thier written texts as basically similar in all langauges. |They are | basically formal debates and legistative in form. | | Keith. The EU MT technologies are specifically adjusted to work with the specialised language/subjects used by the EU for laws and political debates. Googles proposals are for general text, and therefor *much* *much* more demanding. IMO The Google proposals will never get better than a first pass, before a human does the job properly. I use Systran, the market leader, on occasion, and its translations are at best understandable. -- Dave Fawthrop <dave hyphenologist co uk> Freedom of Speech, Expression, Religion, and Democracy are the keys to Civilization, together with legal acceptance of Fundamental Human rights.
Hi There, This debate is becomming very tedious. Am 13.03.2006 um 11:23 schrieb Dave Fawthrop:
On Mon, 13 Mar 2006 10:20:14 +0100, "Keith J. Schultz" <schultzk@uni-trier.de> wrote:
|Maybe one last word here. | | | The EU use MT technologies to translate the bulk. Yet, the produced | texts are still manually processed by humans to get it right. | If the google method was so good the EU would not need translators | since thier written texts as basically similar in all langauges. |They are | basically formal debates and legistative in form. | | Keith.
The EU MT technologies are specifically adjusted to work with the specialised language/subjects used by the EU for laws and political debates.
exactly.
Googles proposals are for general text, and therefor *much* *much* more demanding. More demanding, definately. If it does not even work for a specialized field, how do you expect it to work in a general text!?? I have been here, there and back again.
IMO The Google proposals will never get better than a first pass, before a human does the job properly.
Just what I saying.
I use Systran, the market leader, on occasion, and its translations are at best understandable.
Which product. Already, mentioned that the better products are not availible to the general public. Keith.
On Mon, 13 Mar 2006, Keith J. Schultz wrote:
Hi There,
This debate is becomming very tedious.
Machine Translation is such an important issue that discussion should not be limited, especially here. Thanks!!! Give the world eBooks in 2006!!! Michael S. Hart Founder Project Gutenberg
Am 13.03.2006 um 11:23 schrieb Dave Fawthrop:
On Mon, 13 Mar 2006 10:20:14 +0100, "Keith J. Schultz" <schultzk@uni-trier.de> wrote:
|Maybe one last word here. | | | The EU use MT technologies to translate the bulk. Yet, the produced | texts are still manually processed by humans to get it right. | If the google method was so good the EU would not need translators | since thier written texts as basically similar in all langauges. |They are | basically formal debates and legistative in form. | | Keith.
The EU MT technologies are specifically adjusted to work with the specialised language/subjects used by the EU for laws and political debates.
exactly.
Googles proposals are for general text, and therefor *much* *much* more demanding. More demanding, definately. If it does not even work for a specialized field, how do you expect it to work in a general text!?? I have been here, there and back again.
IMO The Google proposals will never get better than a first pass, before a human does the job properly.
Just what I saying.
I use Systran, the market leader, on occasion, and its translations are at best understandable.
Which product. Already, mentioned that the better products are not availible to the general public.
Keith.
_______________________________________________ gutvol-d mailing list gutvol-d@lists.pglaf.org http://lists.pglaf.org/listinfo.cgi/gutvol-d
On Mon, 2006-03-13 at 10:23 +0000, Dave Fawthrop wrote:
The EU MT technologies are specifically adjusted to work with the specialised language/subjects used by the EU for laws and political debates.
So if Google manages to find a way to classify the domain of a text, it could use domain-specific MT to achieve the same results. I wouldn't be surprised if, using some fairly basic statistic methods, classifying the domain of a text would be trivial. Regards, Walter
participants (4)
-
Dave Fawthrop -
Keith J. Schultz -
Michael Hart -
Walter H. van Holst