I Love Natural Language Processing

[Moses-support] Aligned phrase counts

Hi,

Is there a way to find absolute count (not probabilities) of the number times a phrase is aligned to a target language phrase? I am using the usual arguments to train-factored-phrase-model.perl as suggested on the moses website along with the -debug option, to suppress intermediate file deletion, but don’t see that information anywhere. A look inside train-factored-phrase-model.perl suggested it might have something to do with the phrase_extract/* binaries but I am not sure. Any suggestions/pointers is greatly appreciated.

Thanks,
Delip


Re:[Moses-support] Aligned phrase counts

Hi,

there is no option to report this.

I suggest to change the score.cpp code so that the counts are reported somewhere as they are processed.

Or, sort the extract files and and count up phrase pairs.

-phi

NOTICE:This is digested from the Moses-support mailing list, which supports for the moses SMT decoder.

Related posts:

1. Moses Support Digest:Moses Error in training phrase
2. Moses Support Digest:Hierarchical rule extraction
3. Moses Support Digest:About giza++ options when running moses
4. Moses Support Digest:Translation from English to Foreign Language
5. Moses Support Digest:POS LM
6. Moses Support Digest:Alignment information from binary phrase table
7. Moses Support Digest:Building POS language model with SRILM
8. Moses Support Digest:2.718 in the phrase-table
9. Moses Support Digest: Hierarchical and syntax-based decoding in Moses
10. Moses Support Digest:Reading binary phrase table from the disk