Dear Hieu, I have a tagged corpus so it could easily be made into the format you describe: word1|tag1 word|tag2 etc I've been looking at the Moses Factored Models tutorial (http://www.statmt.org/moses/?n=Moses.FactoredTutorial) and it asks for * surface form * lemma * part of speech * part of speech with additional morphological information Can it be useful to supply Moses with just POS tags as additional information? Thanks, Llio Humphreys
On Fri, Aug 1, 2008 at 1:04 PM, Hieu Hoang <[EMAIL PROTECTED]> wrote: > hi kate > > any tagger is suitable, as long as it produces 1 tag per word (i'm not sure > if any tagger doesn't do this). which tagger you use depend on the language, > eg. i use the TreeTagger for French and the Brill Tagger for English. > > once you have a file with all the tags, and the file with the original text, > use the combine-factors script to create 1 unified file. the unified file > should contain sentences that look like > word1|tag1 word2|tag2 word2|tag3.... > > > -----Original Message----- > From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] > On Behalf Of [EMAIL PROTECTED] > Sent: 31 July 2008 13:52 > To: [email protected] > Subject: [Moses-support] Performing pos tags training > > Hello , my name is Kate , i would like to know what are the steps for > performing training with POS tags, can you recommend me on good tagger which > suitable for the moses system? by searching google , i found insome of the > latest workshops examples for using 2 scripts named : combine-features & > combine-factors . what is thos scripts actions? > _______________________________________________ > Moses-support mailing list > [email protected] > http://mailman.mit.edu/mailman/listinfo/moses-support > > _______________________________________________ > Moses-support mailing list > [email protected] > http://mailman.mit.edu/mailman/listinfo/moses-support > _______________________________________________ Moses-support mailing list [email protected] http://mailman.mit.edu/mailman/listinfo/moses-support
