Computational Linguistics

A discipline in ComputerScience where the focus is linguistics and NaturalLanguage Processing. A yet to be realized goal in this field of study is fully automatic translation between human languages.

20090126 I heard today from VP Research at Google that they are making progress on this front. With 40 languages so far. -- ChrisGarrod see http://translate.google.com/translate_t?hl=en#

Progress maybe, but only one of my three ad-hoc sentences was translated correctly. It seems that everything more complicated than SVO is still very difficult. Phrases and context are handled quite well though,

What about the idea not to aim for natural language processing, but for creation of some sort of simple/pidgin/unknown language which is easy to understand by a computer and a human? Maybe vaguely related to RDF or some other graph representation? Although it surely is disappointing to let aside the aim of creating a truthfully intelligent computer, it could be a very natural step to create a more symbiotic interface between man and machine. Some observations which support this view: Our culture and computers are not distinct spheres anymore, so the computers job isn't necessarily to be as human as possible. People are willing to accept very limited interfaces to write sms, for example. Many many job settings happen in front of a computer. Why not let the computer process patterns of language which it can understand to make the interface more useful? And: A growing part of cultural processes happen in a way that the information can be captured computationally. In the future, social activities like blogging, commenting, tagging and so forth will (probably) be interconnected more densely via the SemanticWeb. What if a crowd-sourcing model could learn from discussions to form Semiotic / semantic webs? Couldn't the large amount of human activities (assessed properly) compensate for the lack of intelligent processing?