Hi,

Currently I am using Moses for one of my research projects for translating
the questionnaires. I have already implemented the base line version of
Moses with Europarl version 7.0 parallel corpus. However, I have got very
low BLEU score which is 13.88.

For tuning, I have used news-test-2015 (French - English) corpus and
testing, I used news-test-2014 (French - English). I used KenLM for the
language model and GIZA++ for word alignment.

I would like to know that how I can tune the translation system further to
improve the translation quality. I would like to know the process of tuning
the machine translation system.

Moreover, previously, I have also tried several times to train UN (French -
English) corpus for the translation system. Unfortunately, every time the
training-model.perl program broke down and could not generate moses.ini in
the ~/working/train/model/ directory. I am using 16 GB RAM and 4 cores
intel machine.



Regarding this issue I would like seek your generous cooperation. Thank.



Regards



*Sohel Mahmud*

*Student Research Assistant*

*Munich Center for the Economics of Aging (MEA) *

*Max Planck Institute for Social Law and Social Policy *

*Amalienstrasse 33*

*80799 Munich, Germany*



*Email mahmud[at]mea.mpisoc.mpg.de <http://mea.mpisoc.mpg.de>*

*Institute homepage **www.mea.mpisoc.mpg.de <http://www.mea.mpisoc.mpg.de/>*
_______________________________________________
Moses-support mailing list
[email protected]
http://mailman.mit.edu/mailman/listinfo/moses-support

Reply via email to