Hello,

My use case is not translation at all, but only in word-aligning
parallel French/English texts within a predefined corpus.

I was able to download and run both MGIZA and pialign on a bitext, but
out of the box this generated abysmal word alignments.

If I understand correctly, what I need to do to improve this is "train a
model".

Unfortunately, I can't find any documentation on how to do this:
everything seems to rush past alignment into machine translation. If
someone could point me in the direction of some docs or ideally
tutorials or examples, it would be very much appreciated.

Presumably I train using the corpus of parallel texts that I want to
eventually word-align between, but I'm not even sure about that at this
point. I haven't been able to get any word aligners other than mgiza and
pialign to run (either they don't compile, or require specialized
preprocessing such as expression trees that I have no idea how to generate).

Many thanks for any help you can provide!

Macho



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

Reply via email to