> Date: Thu, 14 Nov 2013 20:02:23 +0400
> From: Hèctor Alòs i Font <[email protected]>
> To: "[email protected]"
> <[email protected]>
> Cc: Aida Sundetova <[email protected]>,
> "\[apertium-stuff\]" <[email protected]>
> Reply-To: [email protected]
> Subject: Re: [Apertium-stuff] [Apertium-turkic] Guidelines: t1x or t2x?
>
> Mikel, Aida:
>
> It's not a discussion about what is used in t1x and t2x, but in
> http://wiki.apertium.org/wiki/French_and_Esperanto/Notoj_pri_versioj#.C5.9Ctupoj(referencing
>> to the problems explained in
> http://wiki.apertium.org/wiki/Kelkaj_notoj_pri_la_svarmo_de_reguloj )
> there is a discussion about what is done in the n-stage transfer used in
> the French-to-Esperanto translator. In fact, the discussed problem is not
> the limit between t1x and t2x, but what is done the different t(2.n)x
> levels. I'm not sure it may help, but in any case...
>
> Hèctor
>
>
I saw eo-fr translation direction is different from what is done on
other pairs with 8 transfer files.
As Apertium is a research project, I was not surprised something different
was tested for this pair.
If you want to normalise the way different things are done in different
levels of transfer, why not. The problem will be for these things to be
documented.
As some of you may understood, I never did university studies about
automatic translation (or few during last century, and it was more
about compilation) but I learned how Apertium works from the wiki.
During summer 2012, I would have been perfectly unable to work on
transfer rules. Now, the necessary is written on the wiki for the first
level of transfer. There are also texts from Jacob about chunking, but
may be not enough to undestand how 3 levels instead of one may do
things more simple if you need to add chunk tags in level one instead
of directly generating the correct lexical tags for each pattern-item.
But may be students in MT learn more about it in their university.
One thing surprising for me is you propose to do several things
like insterting missing words at the end (4th level in Francis
proposal, third one in Mireia proposal).
The problem is, in the case of an article (determinant) not existing
always in the source language, there will be quite similar rules with
and without the determinant.
I recently proposed to normalize the source language into something
more regular, so at t1x level, less combination of pattern-items
would be needed to process transfer.
I did it in apertium-eo-fr.eo-fr.t0x (in incubator) :
<pattern>
<pattern-item n="nom"/>
</pattern>
<action>
..........
<lit v="un"/>
<lit-tag v="det.ind.sp"/>
..........
<clip pos="1" side="sl" part="whole"/>
..........
</action>
<pattern>
<pattern-item n="det"/>
<pattern-item n="nom"/>
</pattern>
<action>
..........
<clip pos="1" side="sl" part="whole"/>
..........
<clip pos="2" side="sl" part="whole"/>
..........
</action>
Rules of apertium-eo-fr.eo-fr.t1x are still unchanged, so, a lot
of became useless as the patern_item combination will never happen
again after .t0x transfer stage.
But if on the contrary you choose to add missing words at the end
you will need plenty of first level rules for few different variants
in the structure of the sentence.
--------------------------------
Bernard Chardonneau (France)
Phone : [33] 1 64 90 87 04 or [33] 9 72 36 32 90
GSM phone : [33] 6 49 95 13 95
Multilingual websites for my free softwares :
http://libremail.free.fr and http://libremail.tuxfamily.org
http://cyloop.tuxfamily.org (mainly translated with Apertium)
My general website (in french only)
http://bech.free.fr
------------------------------------------------------------------------------
Shape the Mobile Experience: Free Subscription
Software experts and developers: Be at the forefront of tech innovation.
Intel(R) Software Adrenaline delivers strategic insight and game-changing
conversations that shape the rapidly evolving mobile landscape. Sign up now.
http://pubads.g.doubleclick.net/gampad/clk?id=63431311&iu=/4140/ostg.clktrk
_______________________________________________
Apertium-stuff mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/apertium-stuff
