From: Kai Becker (kmb@kai-m-becker.de)
Date: Wed Jan 09 2002 - 02:07:43 MST
Am Mittwoch, 9. Januar 2002 07:36 schrieb Spike Jones:
> In common English speech, for instance, a computer translation
> might puzzle a foreign language speaker if the program did
> not understand the several different meanings of the term
> "fucking," depending on it's position in the sentence.
AFAIK, current machine translation systems already use lists of context
specific translations and rules. The problem is to correctly identify the
semantic context of a given text, which takes more than the rules of
grammar. Is the "green house" a "Treibhaus"(german) for growing plants or a
house painted green? What are white papers then? Are the plants plants
(flowers) or plants (production sites), and what kind of bank is the bank?
"Tampon" has different meanings in printing and personal hygiene, etc.
It gets even worse when the text contains terms of more than one context.
Analogies, jokes, figurative(?) language, and so on.
Kai ("Who the hell is Gen. Failure, and what is he doing with my hard
disk?")
-- == Kai M. Becker == kmb@kai-m-becker.de == Bremen, Germany == "Any technology distinguishable from magic is insufficiently advanced"
This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 09:11:31 MST