the best, easiest way is to use more training data, especially in the
same domain as your test data. And tune your system using data from the
same domain as your test set.
other methods may give you small improvements, but they are
significantly more difficult and language dependent.
On 03/05/2013 16:32, Lingxiao WANG wrote:
Hello,
I want to know, is there any way can help us to improve the
translation from french to chinese, or from chinese to french?
I found some messages in the list, for example, i use the hierarchical
model.
Could you give me some similar suggestions, or some optimization
methods, please?
-------- Message original --------
Sujet: Re: [Moses-support] Does moses supports Chinese ?
Date : Wed, 6 Mar 2013 01:27:43 +0000
De : Philipp Koehn <[email protected]>
Pour : Wang, JinPeng(AWF) <[email protected]>
Copie à : [email protected] <[email protected]>
Hi,
yes, Moses has been used very often for Chinese.
You will get better performance with the hierarchical
model than with the phrase-based model.
One crucial pre-processing step that you will have
to do is word segmentation. There are public
tools available for this task.
Please check public resources such as OPUS
http://opus.lingfil.uu.se/ or the LDC for data.
-phi
best regards,
WANG Lingxiao
_______________________________________________
Moses-support mailing list
[email protected]
http://mailman.mit.edu/mailman/listinfo/moses-support
_______________________________________________
Moses-support mailing list
[email protected]
http://mailman.mit.edu/mailman/listinfo/moses-support
