hi Yanjun,
1. At the moment in the official trunk version of Moses
the word alignment fields are NOT used actually.
2. Phrase table and distortion table are managed
separately by Moses.
best,
Nicola
On Aug 26, 2008, at 11:04 AM, Yanjun Ma wrote:
> Hey,
> Two questions moses decoder:
> 1. The phrase table now has the following format:
> source phrase ||| target phrase ||| word alignment ||| word alignment
> ||| probabilities
>
> The question is: are the entries "word alignment" actually useful to
> be stored in the table since the lexical weighting information has
> been
> compacted in the entry of "probabilities".
>
> 2. If we use the lexical distortion model, does the entries in
> "reordering table" have to be in exactly the same order as the
> "phrase
> table"? In other words, are these two tables are separately stored in
> the decoder implementation?
>
> Thank you very much.
>
> Yanjun
> _______________________________________________
> Moses-support mailing list
> Moses-support@mit.edu
> http://mailman.mit.edu/mailman/listinfo/moses-support
_______________________________________________
Moses-support mailing list
Moses-support@mit.edu
http://mailman.mit.edu/mailman/listinfo/moses-support
