2008/11/12 Felipe Sánchez Martínez <[EMAIL PROTECTED]>:
>
> Hi all,
>

Hi Felipe.

> I am working with an n-best list generated with Moses. My question is,
> which are the individual component scores (unweigthed) obtained?
>
> Suppose the following entry in an n-best list:
>
> 4 ||| así .  ||| d: 0 -0.619042 0 0 0 0 0 lm: -11.4288 tm: -4.84733
> -6.39323 -6.90676 -7.23185 0.999896 w: -2 ||| -2.40665
>
> * "4"
>     -> the number of the sentence
> * "así ."
>     -> the output sentence
> * "d: 0 -0.619042 0 0 0 0 0"
>     -> No idea ¿?
> * "lm: -11.4288"
>     -> I suppose "lm:" stands for language model (log probability).
>        Is this correct?
> * "tm: -4.84733 -6.39323 -6.90676 -7.23185 0.999896"
>      -> I suppose "tm:"  stands for translation model. But,
>         to which translation model correspond each different value?
> * "w: -2"
>      -> I suppose "w:" stands from word penalty.
> * "-2.40665"
>      -> weighted overall socore.
>

http://www.statmt.org/jhuws/?n=Projects.Tuning

"Example: d:1,0.5-1.5 lm:1,0.5-1.5
tm:0.3,0.25-0.75;0.2,0.25-0.75;0.2,0.25-0.75;0.3,0.25-0.75;0,-0.5-0.5
w:0,-0.5-0.5 sets
one weight for the distortion model, starting with 1, then randomized
from 0.5-1.5
one weight for the language model, starting with 1, then randomized from 0.5-1.5
five weights for the translation model:
the first starting at 0.3, then randomized from 0.25-0.75
the first starting at 0.2, then randomized from 0.25-0.75
the first starting at 0.2, then randomized from 0.25-0.75
the first starting at 0.3, then randomized from 0.25-0.75
the first starting at 0, then randomized from -0.5 to 0.5
one weight for the word penalty, starting with 0, then randomized from
-0.5 to 0.5"

>
> I have been looking for this information in  the list archive but my
> search was fruitless.
>

http://www.statmt.org/moses/?n=Moses.AdvancedFeatures
might be useful, too.

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

Reply via email to