Re: [Mt-list] re: need for Englsih corpus to MT evaluation

[ESTeam AB]

 Dear Mr. Arabsorkhi,   there is no purpose in evaluating MT systems like you propose. Its purpose is to translate not to analyse language and according to some schools i.e. most of the commercial systems, the two simply have very little to do with the other. Most translation problems are lexical i.e. ambiguities in the same category and analysis hasn't to date provided any solutions. If I was you I would pick a better research topic.   Sorry to be so blunt, but I don't like good minds to waste their time on things in MT that do not make the field evolve and improve.   Kind Regards, Gudrun Magnusdottir, CEO ESTeam AB   www.esteam.gr    ----- Original Message ----- From: Mohsen Arabsorkhi To: mt-list@eamt.org Sent: Thursday, May 25, 2006 9:07 AM Subject: [Mt-list] re: need for Englsih corpus to MT evaluation Hi friends,   First I want to thank for your help. I think I couldn't describe my problem. I research about Standard Evaluation of MT system. So I want to know how  the quality of system is on Morphological Analysis, Parsing, Translation, and Generation. In the field of Parsing, I need a corpus contains sentences that cover most English syntactic rules (grammar). So it should not be from specific domain or limited to some constraint, it must be really general and comprehensive. using this corpus I will test the system and evaluate parse trees to find out the MT quality in covering syntax.   Best regards, Mohsen Arabsorkhi _______________________________________________ Mt-list mailing list

_______________________________________________
Mt-list mailing list