Dear João Graça and list suscribers,
Fields 3 and 4 are word alignment information: which words from the e
phrase align to words of the f phrase. You might want to consult the
following thread:
http://thread.gmane.org/gmane.comp.nlp.moses.user/510/focus=511
I once used the following command line in order to get rid of those
fields and keep the translation table smaller:
cat phrasetable.tt | sed 's/|||/~/g' | awk -F "~" '{ print $1 "|||" $2
"|||" $5}' > filteredtable.tt
I replace ||| through ~ (as '~' did not appear at all in my table) and
then use awk to recreate the table and filter out fields 3 and 4.
In a few cases, it didn't produce valid lines, I had to correct some
lines of the table. It's very basic so if someone has improvement
suggestions, these are welcome.
Emmanuel
On Thu, Apr 10, 2008 at 3:56 AM, João Graça <[EMAIL PROTECTED]> wrote:
> Hello,
>
> I have a question regarding the phrase table fields:
>
> Given the following entry from the phrase table:
>
> è ||| will be ||| (1) ||| () (0) ||| 0.0142857 0.0067227 0.000222767
> 5.15661e-06 2.718
>
> What does the 3 and 4 entry mean? The description for this entries is not
> present at the moses home page.
>
> Thanks a lot,
>
> João Graça
>
> _______________________________________________
> 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
