Hi,

Ok this seems reasonable. In the example I gave there are 5 unknown
words, thus the number of phrases is 17 as Philipp said.

Thank you very much for your help.

--
Felipe

El mié, 16-09-2009 a las 16:48 +0100, Hieu Hoang escribió:
> it may be an unknown word issue. If a word is unknown, it doesn't have 
> translation model scores.
> 
> there may be a few unknown words for this particular sentence as its 
> full of numbers and code
> 
> Felipe Sánchez Martínez wrote:
> > Hi Philip,
> >
> > Notice that in the example I gave the number of phrase pairs used is 22.
> >
> > --
> > Felipe.
> >
> > El mié, 16-09-2009 a las 16:24 +0100, Philipp Koehn escribió:
> >   
> >> Hi,
> >>
> >> the phrase penalty is part of the translation model - in a very
> >> crude way: each phrase pair entry has a 5th scoring component
> >> which is 2.76, which is almost e. Hence if there are 17
> >> phrase pairs used, the log score is 16.9982 (log 2.76^17).
> >>
> >> -phi
> >>
> >> 2009/9/16 Felipe Sánchez Martínez <[email protected]>
> >>         Hi there,
> >>         
> >>         How is calculated the phrase penalty? In my n-best list I have
> >>         the
> >>         following hypothesis (I have used # to divide the translation
> >>         in the
> >>         phrases used, maybe it helps)
> >>         
> >>         1 ||| el siguiente punto es # la recomendación para la segunda
> >>         lectura
> >>         ( #a4-0294 # / 96 # ) en nombre de la comisión # de
> >>         transportes y
> >>         turismo , # sobre la posición común adoptada por el # consejo
> >>         # con
> >>         vistas a la adopción de # una directiva del consejo sobre # el
> >>         equipamiento marino # ( # c4-0370 # / # 96-95 # / # 0163 # ( #
> >>         syn # ) )
> >>         ( ponente : # sr. kaklamanis # ) .||| d: 0 -7.54513 0 0
> >>         -5.7395 0 0 lm:
> >>         -353.572 tm: -3.94358 -37.6388 -5.04752 -57.4153 16.9982 w:
> >>         -64
> >>         
> >>         According to this the phrase penalty is 16.9982. How is this
> >>         value
> >>         calculated?
> >>         
> >>         Thank you very much in advance
> >>         Regards
> >>         --
> >>         Felipe
> >>         
> >>         _______________________________________________
> >>         Moses-support mailing list
> >>         [email protected]
> >>         http://mailman.mit.edu/mailman/listinfo/moses-support
> >>
> >>     
-- 
Felipe Sánchez Martínez <[email protected]>
Departamento de Lenguajes y Sistemas Informáticos
Universidad de Alicante, E-03071 Alicante (Spain)
Tel.: +34 965 903 400, ext: 2966 Fax: +34 965 909 326
http://www.dlsi.ua.es/~fsanchez

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

Reply via email to