Hi,

>  Another question that i have been thinking is:
>
>  1. Can we  have a 'n-lingual corpus' using the statistical approach?
>
>  I mean can we have the source along with its translation in multiple
>  languages and then train the parallel data. Is this really possible?

The Europarl corpus is to a large degree such a corpus
http://www.statmt.org/europarl/

>  2. If we have a bi-lingual corpus, is it easy to interchange source with
>  target or will have to create another parallel corpus?

We do this all the time...

-phi
_______________________________________________
Moses-support mailing list
[email protected]
http://mailman.mit.edu/mailman/listinfo/moses-support

Reply via email to