Thanks Hieu. Let me tell you what I am trying. I am getting the output of ASR system in confusion networks format which has words with its probability. I am converting the confusion network in the word lattice format which will be input to the moses. Can the confusion network probabilities be used for these lattices? Also how does moses decode these word lattices?
Regards, Sanket Gandhare On 21 August 2017 at 02:49, Hieu Hoang <hieuho...@gmail.com> wrote: > the probabilities are calculated by the external tool that creates the > lattice, eg. the segmentation tool may give probabilities for the different > segmentations it makes. The probabilities are used as a feature in the > log-linear model in the same way as probabilities from the phrase-table or > language model etc > > > On 20/08/2017 13:39, Sanket Gandhare wrote: > > Hello all, > > I want to give input to moses in lattice format but I am not sure what > the probabilities in the lattice mean. I went through this link : > > http://www.statmt.org/moses/?n=Moses.WordLattices > > Also, how these probabilities calculated. What algorithm Moses uses to > decode these word lattices? > > Please give me idea of all this. > > > _______________________________________________ > Moses-support mailing > listMoses-support@mit.eduhttp://mailman.mit.edu/mailman/listinfo/moses-support > > > -- > Hieu Hoanghttp://moses-smt.org/ > >
_______________________________________________ Moses-support mailing list Moses-support@mit.edu http://mailman.mit.edu/mailman/listinfo/moses-support