On 14 May 2011 00:47, Paulo Schreiner <[email protected]> wrote:
> Em Sex, 2011-05-13 às 23:45 +0100, Jimmy O'Regan escreveu:
>> On 13 May 2011 22:55, Paulo Schreiner <[email protected]> wrote:
>> > Anyone here has some experience with the apertium tagger?
>> >
>> > I have created (to my best knowledge) all required resources, but got
>> > stuck with the following error:
>> >
>> > apertium-tagger -d -s 0 pt.expand pt.tagged.txt pt.tsx pt.prob pt.tagged
>> > pt.tagged.morf
>> > Calculating ambiguity classes...
>> >
>> > 30 states and 31 ambiguity classes
>> > Kupiec's initialization of transition and emission probabilities...
>> > Initializing transition and emission probabilities from a hand-tagged
>> > corpus...
>> > {adv}    Word: depois -- {prp,adv}       Word: depois
>> > Error: A new ambiguity class was found. I cannot continue.
>> > Word 'depois' not found in the dictionary.
>> > New ambiguity class: {prp,adv}
>> > Take a look at the dictionary, then retrain.
>>
>> 'depois' needs to be added to the dictionary (as both preposition and
>> adverb), to match the corpus. In all likelihood, the word is present
>> (otherwise it couldn't have encountered an ambiguity), so you'll
>> probably need to look at the commands in the Makefile that are used to
>> filter the output of lt-expand - it's discarding too much.
>>
>
> Like this? I sorted the expanded file, seems they are there.
>
> depois:depois<adv>
> Depois:depois<adv>
> depois:depois<prp>
> Depois:depois<prp>
>
> Any other idea?

No need for another idea, because I'm right :P

That's the wrong format. It should match the output of the analyser
(i.e., you should have entries like:
^depois/depois<pr>/depois<adv>$
instead of what you have).

-- 
<Sefam> Are any of the mentors around?
<jimregan> yes, they're the ones trolling you

------------------------------------------------------------------------------
Achieve unprecedented app performance and reliability
What every C/C++ and Fortran developer should know.
Learn how Intel has extended the reach of its next-generation tools
to help boost performance applications - inlcuding clusters.
http://p.sf.net/sfu/intel-dev2devmay
_______________________________________________
Apertium-stuff mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/apertium-stuff

Reply via email to