Hi Nurendra,
2015-03-01 6:18 GMT+01:00 Nurendra Choudhary <[email protected]>
:
> Hi Mikel,
>
> Shouldn't echo "बागेत" | apertium-destxt | lt-proc mar-hin.automorf.bin
> give ^word/lemma<morphological analysis>$ [
> http://wiki.apertium.org/wiki/Contributing_to_an_existing_pair#The_morphological_analyser_output
> ]
>
That's right, and that's what your morphological analyzer does.
Generally speaking, it will output ^surfaceForm/lemma<tags>$. Lemma + tags
is what is called a 'lexical form'.
> ^बागेत/बाग<n><f><sg><loc>$^./.<sent>$[][ (This should be correct
> then, right?)
>
I don't speak Marathi, so obviously can't tell whether the analysis is
correct, but here बागेत is the surface form, and बाग is the lemma.
In some cases, multiple analyses of a surface form are possible. It would
like this: ^surfaceForm/lemma1<tags1>/lemma2<tags2>/...$.
Tagger selects one lexical form it thinks is correct/most probable. That
means that from that point on you don't have surface forms in the pipeline,
you only have lemmas. Assuming that the tagger selects the first lexical
form (the first "reading"), it would look like this: ^lemma1<tags1>$.
In your example that's ^बाग<n><f><sg><loc>$.
But your bidix entry <e><p><l>बागेत<s n="n"/><s n="f"/></l>
<r>बाग<b/>में<s n="n"/><s> n="m"/></r></p></e> expects बागेत for the
left-hand side (Marathi) *lemma*, not बाग. That is why it doesn't get
translated.
You might find apertium-viewer [1] or apertium-view.sh [2] helpful to
easily see/modify how text passes through the whole pipeline.
[1] http://wiki.apertium.org/wiki/Apertium-viewer
[2] http://wiki.apertium.org/wiki/Apertium-view.sh
Best,
Ilnar
I didn't completely understand the point. Can you please elaborate it a
> little?
> Thanks for looking into it.
>
> -Nurendra Choudhary.
>
>
> On Sat, Feb 28, 2015 at 12:07 PM, Mikel Forcada <[email protected]> wrote:
>
>> Hi Nurendra,
>>
>> clearly, your form "बाग<n><f><sg><loc>" starts with "बाग<n><f>" and not
>> with "बागेत<n><f>" as your dictionary entry expects. Therefore, the
>> dictionary entry you show does not match it.
>>
>> Hope this helps
>>
>> Mikel
>>
>>
>>
>>
>> EL Sat, 28 Feb 2015 06:15:53 +0530
>> Nurendra Choudhary <[email protected]> escrigué:
>>
>> > Hi,
>> >
>> > I wanted to know how apertium handles inflections or the bilingual
>> > dictionary.
>> > This is the entry in the bidix:
>> > <e><p><l>बागेत<s n="n"/><s n="f"/></l> <r>बाग<b/>में<s n="n"/><s
>> > n="m"/></r></p></e>
>> > बागेत = बाग(lemma) + ेत [handled in the marathi monodix]
>> > Also बाग, में are both part of the hindi monodix.
>> >
>> > but echo "बागेत" | apertium -d . mar-hin
>> > gives : बाग
>> > expected by me : बाग में
>> >
>> > echo "बागेत" | apertium-destxt | lt-proc mar-hin.automorf.bin
>> > gives ^बागेत/"$^./.<sent>$[][ <This is perfect>
>> >
>> > But, I think the error maybe here,
>> > echo "बागेत" | apertium-destxt | lt-proc mar-hin.automorf.bin |
>> > apertium-tagger -g mar-hin.prob
>> > ^बाग<n><f><sg><loc>$^.<sent>$[][ <There is no बागेत here meaning it
>> > won't be moving forward in the pipeline>
>> >
>> > Can anybody please tell me where I am going wrong?
>> >
>> > - Nurendra Choudhary.
>>
>>
>>
>> --
>> Mikel L. Forcada <[email protected]>
>> http://www.dlsi.ua.es/~mlf/
>> +34 96 590 9776
>> Departament de Llenguatges i Sistemes Informàtics
>> Universitat d'Alacant
>> E-03071 Alacant (Spain)
>>
>>
>> ------------------------------------------------------------------------------
>> Dive into the World of Parallel Programming The Go Parallel Website,
>> sponsored
>> by Intel and developed in partnership with Slashdot Media, is your hub
>> for all
>> things parallel software development, from weekly thought leadership
>> blogs to
>> news, videos, case studies, tutorials and more. Take a look and join the
>> conversation now. http://goparallel.sourceforge.net/
>> _______________________________________________
>> Apertium-stuff mailing list
>> [email protected]
>> https://lists.sourceforge.net/lists/listinfo/apertium-stuff
>>
>
>
>
> ------------------------------------------------------------------------------
> Dive into the World of Parallel Programming The Go Parallel Website,
> sponsored
> by Intel and developed in partnership with Slashdot Media, is your hub for
> all
> things parallel software development, from weekly thought leadership blogs
> to
> news, videos, case studies, tutorials and more. Take a look and join the
> conversation now. http://goparallel.sourceforge.net/
> _______________________________________________
> Apertium-stuff mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/apertium-stuff
>
>
------------------------------------------------------------------------------
Dive into the World of Parallel Programming The Go Parallel Website, sponsored
by Intel and developed in partnership with Slashdot Media, is your hub for all
things parallel software development, from weekly thought leadership blogs to
news, videos, case studies, tutorials and more. Take a look and join the
conversation now. http://goparallel.sourceforge.net/
_______________________________________________
Apertium-stuff mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/apertium-stuff
