Hi Simon If you look at the 'extract' file that Moses creates during training, it just contains a raw list of all the extracted phrase pairs. If you want to find the most common ngram pairs then you could get them from this file.
In the phrase table, the phrase counts are normalised to give p(e|f) and p(e|f). cheers - Barry Quoting Simon Hafner <reactorm...@gmail.com> on Sun, 3 Jun 2012 21:26:28 +0200: > Hi all > > is there a nice way to get the top 100 translations? > > I'm trying to compare two languages on character ngram level, to find > common edit paths. The idea is to train moses for that pair and then > extract the most common ngram pairs. Is this even possible or are they > normalized based on their occurrence? > > -- Simon > _______________________________________________ > Moses-support mailing list > Moses-support@mit.edu > http://mailman.mit.edu/mailman/listinfo/moses-support > > -- The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336. _______________________________________________ Moses-support mailing list Moses-support@mit.edu http://mailman.mit.edu/mailman/listinfo/moses-support
