Talk:Eurotra
This article is rated Start-class on Wikipedia's content assessment scale. It is of interest to the following WikiProjects: | ||||||||||||||||||||||||
|
AndyHolyer's recollections of the Eurotra project
[edit]I worked on Eurotra from 1986 to late 1988: my memories of technical details may be a bit sloppy (Id welcome help with this). I'll try to expand the entry in the next few weeks.
- AndyHolyer 16:40, Jun 2, 2005 (UTC)
- Hi, technical information isn't important and may even be a hindrance. Try to picture a reader intelligent and well-informed (she arrived at this article for a reason) but neither a computer scientist nor a computational linguist. Technical descriptions belong in offline references, of which there must be various. HTH -- Viajero | Talk 17:47, 2 Jun 2005 (UTC)
The big mystery
[edit]Hi Andy,
What I'd LOVE to know -and this piece of info could appear in the article- is how come all of you linguists at the then called European Community could not realize (in the 14 years that the project lasted) that in order to come up with a translation algo, all you had to do is use the basic principles of compared syntax,in this case establish patterns between pairs of languages.
Example of pattern:
A white dog ART CADJ NS
=
un chien blanc ART NMS CADJ
This is certainly not rocket science but still no one working on the Eurotra project ever came up with this basic algo and it took engineers at Google and Microsoft a good 10 years to come up with it.
I just don't get it!
-- Robert Abitbol — Preceding unsigned comment added by 24.54.3.100 (talk) 18:05, 11 December 2019 (UTC)