More comments below:

Could you give more details of what was done with the Hebrew analyser
generated from Hspell ? E.g. which categories have been generated, how
many entries, what coverage etc. Also, do you think you will have time
to add the verbs ? 

I think perhaps it would be worth noting that the pair will be
unidirectional (Maltese→Hebrew).        

I would be interested in seeing some small evaluation of existing MT
systems for mt-he (e.g. Google translate) and their strong points and
weaknesses... How do you think a rule-based system based on Apertium
might compare ? 

Could you justify the text. I think it would look nicer in LaTeX, but if
Libre/Open/...Office is your thing, then that's ok.

I'll ask again, will you use separate frequency lists and then take the
intersection via the bilingual dictionary ? Or will you just use one
list ? 

You have allocated three weeks to making bilingual dictionary entries.
How many entries do you think you can do per hour ? How many does this
add up to over this period ? 

Again, no info on the parallel corpora. The Bible is usually a popular
source of parallel text -- although very restricted domain -- but do
they bother to translate the whole Bible into Hebrew ?

ex.

Fil-bidu Alla halaq is-sema u l-art
בראשית ברא אלהים את השמים ואת הארץ 
Bereishit bara Elohim et hashamayim ve'et ha'aretz. 

Fil-bidu = Bereishit
Alla     = Elohim
halaq    = bara
sema     = shamayim
art      = aretz

etc.

What do you consider to be "Desirable coverage" ? 

Thanks again!

Fran

El dl 04 de 04 de 2011 a les 00:39 +0300, en/na Sagie Maoz va escriure:
> Hi,
> 
> 
> Thanks Francis for the comments and guidance.
> 
> 
> I have updated my application draft for my project with regards to
> most of your notes.
> I am still unclear about the implications of using hand-tagged corpora
> for both languages, and would love a clarification. For my work plan,
> I used past years' applications as reference.
> 
> 
> I have also added a part about my research and collected resources,
> especially in regards to morphological analysers for both languages.
> 
> 
> Please find my application available
> at http://sagie.maoz.info/apertium-application.pdf.
> It would be great to get more comments or suggestions from anyone.
> 
> 
> Thanks,
> 
> On Sun, Apr 3, 2011 at 3:47 AM, Francis Tyers <[email protected]>
> wrote:
>         El dg 03 de 04 de 2011 a les 01:15 +0300, en/na Sagie Maoz va
>         escriure:
>         > Dear All,
>         >
>         >
>         > I'm happy to attach my application draft for a GSoC project,
>         a
>         > Maltese-Hebrew language pair.
>         > I will be happy to hear your thoughts and comments,
>         particularly on
>         > when should I submit it at the Google Melange site.
>         >
>         >
>         > Please find my application available
>         > at http://sagie.maoz.info/apertium-application.pdf.
>         
>         
>         Hi, nice application... I think you are underselling yourself
>         though.
>         I've chatted with you a bit on IRC, and I think you have
>         already
>         gathered quite a lot of references, possible sources of
>         information for
>         both languages. So it might be a good idea to have a section
>         in your
>         proposal "Resources" (and include here grammars, dictionaries
>         etc.)
>         
>         If you are able to get access to the morphological analyser
>         for Hebrew
>         before the 8th it should be included in your application. If
>         you are not
>         able to, then it should be assumed that it is _not_ available,
>         and the
>         time plan should be adjusted accordingly. Note that "promised
>         access to"
>         is not quite the same as "will be released under the GPL" --
>         which would
>         be our requirement for using it.
>         
>         I'm not sure if a morphological analyser for Maltese exists,
>         but if it
>         does, then the same goes for that as for the Hebrew one.
>         
>         I'm not sure about the "Renaissance of Hebrew and Maltese"
>         reference.
>         Perhaps there is a more scholarly reference available ?
>         
>         * bellow -> below
>         
>         Where are you expecting to find a parallel corpus for Hebrew
>         and
>         Maltese ? One person you might try contacting on this question
>         is Adam
>         Ussishkin, who I met at SEPLN and is a nice guy.
>         
>         I'm missing a bit of a description of the languages, and some
>         potential
>         transfer problems. I'm not expecting you to have done all the
>         work, but
>         some examples would be nice :) Perhaps you could demonstrate
>         by
>         translating the examples from
>         http://wiki.apertium.eu/index.php/Appendix_A:_Frequency as a
>         computer
>         would ?
>         
>         ==Week plan==
>         
>         * Are you going to be using separate frequency lists for
>         Hebrew and
>          Maltese ? If so, then might you not end up with having to
>         take the
>          intersection and losing words -- thus not completely
>         effectively using
>          your time ?
>         
>         * Another idea might be to just use a frequency list from the
>         Maltese
>          side ?
>         
>         * I think you are optimistic about getting useful handtagged
>         corpora for
>          both Maltese and Hebrew -- in order to include them in your
>         proposal,
>          you should have them _before_ the 8th April. Otherwise assume
>         that
>          they don't exist... remember Zigglebottom.
>         
>         * Even if you do get hand tagged corpora, you would need to
>         convert them
>          to the tagset of your analyser(s), and also you might need to
>         fix
>          tokenisation issues.
>         
>         * I think that a week for working on the bilingual dictionary
>         is too
>          little. When working from a language that you don't know into
>         one
>          that you do know, you need to be particularly careful. For
>          Macedonian--English, it took me approx 8 days. And I have
>         quite a lot
>          of experience.
>         http://permalink.gmane.org/gmane.comp.nlp.apertium/285
>         
>         The thanks in Maltese/Hebrew is a really nice touch :)
>         
>         Fran
>         
>         
>         
> ------------------------------------------------------------------------------
>         Create and publish websites with WebMatrix
>         Use the most popular FREE web apps or write code yourself;
>         WebMatrix provides all the features you need to develop and
>         publish your website. http://p.sf.net/sfu/ms-webmatrix-sf
>         _______________________________________________
>         Apertium-stuff mailing list
>         [email protected]
>         https://lists.sourceforge.net/lists/listinfo/apertium-stuff
> 
> 
> 
> -- 
> Your friend in time,
> Sagie Maoz
> [email protected] // +1 (347) 556.5044 // +972 (52) 834-3339
> http://sagie.maoz.info/ http://n0nick.net/
> 
> /* simba says roar! */
> 
> 
> ------------------------------------------------------------------------------
> Create and publish websites with WebMatrix
> Use the most popular FREE web apps or write code yourself; 
> WebMatrix provides all the features you need to develop and 
> publish your website. http://p.sf.net/sfu/ms-webmatrix-sf
> _______________________________________________
> Apertium-stuff mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/apertium-stuff



------------------------------------------------------------------------------
Create and publish websites with WebMatrix
Use the most popular FREE web apps or write code yourself; 
WebMatrix provides all the features you need to develop and 
publish your website. http://p.sf.net/sfu/ms-webmatrix-sf
_______________________________________________
Apertium-stuff mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/apertium-stuff

Reply via email to