Hi all,

Is there a way that I can take a model, a source sentence S, a desired
target sentence T, and have Moses give me the best scores associated
with selecting T as a translation? I'm most interested in the p(f|e)
for T and S.

For example, let's say I have a source sentence in both French and
German, and English is my target. If the fr-en model returns a
translation, I'd like to know what score the de-en model would give
that translation (given the equivalent German source).

I tried using the -xml-input flag to force Moses to choose the
translation that I want, but when I did that the result from Moses
doesn't appear to be using the actual weights from the phrase table -
p(f|e) is zero in the result returned by Moses.

Is there a way to find out what score Moses would give to a specific
target sentence?

Thanks,
Lane Schwartz

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

Reply via email to