Hi anyone
I've added optional extra information as the 6th column in each rule of
the SCFG phrase table. You can get it by using the argument
--OutputNTLengths
in the extract, score & consolidate programs.
It tell you the distribution of words covered by each non-terminal in
the rule when they were extracted from the corpus. eg.
2|S|2=0.4
means the 40% of non-terminals in the 2nd position spans 2 words in the
source sentence.
If you want to use it, feel free. Just let me know
_______________________________________________
Moses-support mailing list
[email protected]
http://mailman.mit.edu/mailman/listinfo/moses-support
