Dear Apertiumers:
See attached. The eng-kaz language pair may be ready to move to staging,
and the group at the Kazakh National University think it should be done.
If this is the case, could someone who has already done this please
explain how to do it, and answer Zhandos's questions below?
1. What do we need to do in order to start moving the pair?
2. And how moving will affect working with the repository?
3. Will we have to stop doing anything with it for a while?
4. When would that happen if it will happen?
I would appreciate it very much if someone could help us.
Thanks a million.
Best,
Mikel
-------- Missatge reenviat --------
Assumpte: Moving eng-kaz to staging
Data: Thu, 12 Nov 2015 14:13:26 +0600
De: Zhandos Zhumanov <[email protected]>
A: Mikel L. Forcada <[email protected]>, Ualsher Tukeyev
<[email protected]>
CC: Aida Sundetova <[email protected]>, Diana Rakhimova
<[email protected]>
Dear Mikel,
We have checked if eng-kaz pair meets requirements for staging that are
written here: http://wiki.apertium.org/wiki/Staging - the result of our
"inspection" is in the ISSUES file that Aida uploaded in the pair folder
(that is one of the requirements). I also attached it to this message.
We start working on cleaning errors and crashes.
We still have some questions: What do we need to do in order to start
moving the pair? Do we apply somewhere or notice someone besides you?
And how moving will affect working with the repository? Will we have to
stop doing anything with it for a while? And when would that happen if
it will happen?
Best Regards,
Zhandos Zhumanov
2015-11-10 8:26 GMT+06:00 Ualsher Tukeyev <[email protected]
<mailto:[email protected]>>:
Dear Mikel,
We are agree with your proposals, these are really situation.
About publications. We will discuss a branch of publications
preparing and will tell you.
Best regards
Ualsher Tukeyev
email: [email protected] <mailto:[email protected]>
tel. 7-727-3381077 <tel:7-727-3381077>
mob. 7-701-7106351 <tel:7-701-7106351>
2015-11-05 22:19 GMT+06:00 Mikel L. Forcada <[email protected]
<mailto:[email protected]>>:
Dear Ualsher,
I just realised I had not answered this message.
I am adding Aida, as I know she is on top of the eng-kaz
language pair.
(1) Regarding moving eng-kaz from the incubator up, Francis
Tyers and I think it should move to "staging" [1], as currently
the system still has about 5% generation errors going from eng
to kaz and some segmentation faults (crashes) and about 15%
generation errors when going to kaz to eng. Please let us know
if we agree and we will do that.
For the pair to be moved to stable, there should be virtually no
generation errors, and no segfaults. I would encourage your team
to fix these errors by diagnosing them on a large corpus (e.g.
Wikipedia dumps). We could Miquel could help to arrange corpora
and tests.
(2) Regarding moving rus-kaz, it could be moved to "nursery"
[2], but no further. We are not sure if you have been working in
both directions, but here are samples of output:
kaz->rus
землю #Сейчас Казахстан сын давнее человечество из этого *1
#миллион
поселение раньше года *еткен. Бронза в веке *Сібірдің, *Жайық
местности,
Казахстан я #юноша #Средний Азии-*ғайыр поля *тегі и #исторический
судеб *ортақтығы с родственным боком племена *мекендеді. Эти
племена сам
в одной модели светлая культура оставил. Это в науке *Андрон
культура
*деп *атайды. #век #Ранний #железный #век (из нашей эры #раньше #I
*мыңжылдық #среда) в Казахстане *тайпалық союзы *қалыптасты.
Казахстана
южного, выход и центральные районы *мекендеген племена союзу
осторожные
племена, а западного, племена в северных районах *савроматтар
*тайпалық
объединению *бірікті.
rus->kaz
@литовский мемлекеттің @Зарождение @относить @к X—XIII #дәуір. Ол
*формировалось территорияда екі #дардай *балтских тайпалардың:
*Жемайтии @CNP> *Аукштайтии. @Уже @Великий @княжество құрамында
@Литовский *Жемайтия @CNP> *Аукштайтия *унаследовали #аталым
@другой,
@этнический #туысқан @к олар, *балтского @Литва тайпасына, @который
@жило территорияда @современный #Беларуссия ауданда Сыртқы
*Понеманья
@между *Слонимом, @Минский, *Молодечном мен *Новогрудком.
There are far too many diagnostics (@, #) and many unknown
words. I would keep it in incubator at least until the number of
diagnostics and unknown words is drastically reduced.
(3) On a substantial publication (a journal paper, I think you
mean), I don't know enough about your work during 2015 (I have
to read the Russian report) but as regards our part, there is
not much to publish yet: most of the work has been of a
preparatory nature and feasibility studies. For instance, to
really check the eng-kaz rule generation, we would probably need
a larger parallel corpus and a more stable language pair with
less diagnostics. Please let me know what you think.
Best regards,
Mikel
[1] http://wiki.apertium.org/wiki/Staging
[2] http://wiki.apertium.org/wiki/Nursery
2015-10-21 18:08 GMT+02:00 Ualsher Tukeyev
<[email protected] <mailto:[email protected]>>:
Dear Mikel,
in report which we sent to Ministry, we attach all your
reports in application B.
But I not attach your reports to file sended to you, because
you wrote and have these reports.
No, I not need additional input from you.
Thanks you and your team for substantial contribution to our
join report.
Sorry for Russian language of report , if it is need some
parts we can translate on English.
Now we continue to improve of current results: we will
increase of number words in dictionaries, number of created
rules. For this we will use all created by your team tools,
particular, last sended tools(Spiderling and tool for
addition word to dictionary). Here we will ask consultation
of Miquel.
Also, Mikel, we have to discuss about join substantial
publication on results of this year. You and your team have
reach experience in this.
Also, I ask members of my team to make estimation of each
pair of languages by Apertium methodology.
We will ask you to move some pair from Incubator to more
high level by results of estimation. We will send you that
information of estimation and our proposal what pair in what
level deserve(may be some pair left in incubator).
Of course, I hope that our report will be supported to
continue financing on next year.
Best regards
Ualsher Tukeyev
email: [email protected]
<mailto:[email protected]>
tel. 7-727-3381077 <tel:7-727-3381077>
mob. 7-701-7106351 <tel:7-701-7106351>
2015-10-21 20:31 GMT+06:00 Mikel Forcada <[email protected]
<mailto:[email protected]>>:
EL Wed, 21 Oct 2015 18:35:11 +0600
Ualsher Tukeyev <[email protected]
<mailto:[email protected]>> escrigué:
> Dear Mikel,
>
> we have completed our report, see attached file.
> Yours reports we include to application B, supported
first page,
> where writed staff from your team and in what chapter
in Russian
> include what your report.
>
> Best regards
>
> Ualsher Tukeyev
> email: [email protected]
<mailto:[email protected]>
> tel. 7-727-3381077 <tel:7-727-3381077>
> mob. 7-701-7106351 <tel:7-701-7106351>
Dear Ualsher,
I am not completely sure what you mean in your sentence:
"Yours reports
we include to application B, supported first page, where
writed staff
from your team and in what chapter in Russian include
what your
report.".
If you need some additional input from us, please let me
know as soon
as possible, and we will also prepare it as soon as
possible.
Thanks a lot for the final effort and also for the great
opportunity you
have given us through this project. I hope the project
can have a
successful termination!
Best regards,
Mikel
--
Mikel L. Forcada <[email protected] <mailto:[email protected]>>
http://www.dlsi.ua.es/~mlf/ <http://www.dlsi.ua.es/%7Emlf/>
+34 96 590 9776 <tel:%2B34%2096%20590%209776>
Departament de Llenguatges i Sistemes Informàtics
Universitat d'Alacant
E-03071 Alacant (Spain)
--
Mikel L. Forcada E-mail: [email protected] <mailto:[email protected]>
Departament de Llenguatges Phone: +34-96-590-9776
<tel:%2B34-96-590-9776>
i Sistemes Informàtics also +34-96-590-3772
<tel:%2B34-96-590-3772>.
UNIVERSITAT D'ALACANT Fax: +34-96-590-9326
<tel:%2B34-96-590-9326>, -3464
E-03071 ALACANT, Spain.
URL: http://www.dlsi.ua.es/~mlf <http://www.dlsi.ua.es/%7Emlf>
------------------------------------------------------------------------------
_______________________________________________
Apertium-stuff mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/apertium-stuff
