Difference between revisions of "Turkish and Kyrgyz/Kymorph article"

From Apertium
Jump to navigation Jump to search
(Undo revision 28838 by Firespeaker (Talk))
 
(14 intermediate revisions by 2 users not shown)
Line 1: Line 1:
* Outline
== Outline ==


== General background ==
⚫
* Morphotactica
* Submitting abstract to: [http://www.lrec-conf.org/lrec2012/ LREC 2012 Istanbul]
* Deadline: October 15, 2011


== Similar articles ==
⚫
* Morphophonologia
* [http://www.mt-archive.info/FreeRBMT-2009-Faridee.pdf Abu Zaher Md. Faridee & Francis M. Tyers - Development of a morphological analyser for Bengali]
* [http://www.let.rug.nl/coltekin/papers/coltekin-lrec2010.pdf Çagrı Çöltekin - A Freely Available Morphological Analyzer for Turkish]


⚫
== Morphotactica ==
⚫
* Corpora
* Irregular negatives of many verb forms


⚫
== Morphophonologia ==
⚫
* Numbers
* /рн/ nouns

⚫
== Corpora ==
* Which corpora to use?
** Wikipedia
**# [http://paste.pocoo.org/show/2KXepqcTTiWDWLOFPRlR/ punktgen.py] ky.crp.txt ky.pickle
**# aq-wikicrp -x -t ky.pickle [http://download.wikimedia.org/kywiki/20110923/kywiki-20110923-pages-articles.xml.bz2 kywiki-20110923-pages-articles.xml] kywp.xml
** [[Turkish and Kyrgyz/Making a corpus from azattyk|Azattyk]]
* concerns
** Wikipedia is messy; should we have an automated cleaning process or get stats as-is?
*** Use aq-wikicrp, this way it is reproducible .

⚫
== Numbers ==
{|class="wikitable"
|+ size of corpora
|-
|
! wikipedia
! azattyk 2010
! all azattyk
|-
! num articles
| 1531([http://ky.wikipedia.org/wiki/Special:Statistics ?], [http://dumps.wikimedia.org/kywiki/20110923/ ?])
| 9803 (6627?)
|
|-
! num words
| 271005
| 3394686
|
|-
! xml file size
| 3.8MB
| 49MB
|
|}

Latest revision as of 16:50, 13 October 2011

Outline[edit]

General background[edit]

Similar articles[edit]

Morphotactica[edit]

  • Irregular negatives of many verb forms

Morphophonologia[edit]

  • /рн/ nouns

Corpora[edit]

  • Which corpora to use?
  • concerns
    • Wikipedia is messy; should we have an automated cleaning process or get stats as-is?
      • Use aq-wikicrp, this way it is reproducible .

Numbers[edit]

size of corpora
wikipedia azattyk 2010 all azattyk
num articles 1531(?, ?) 9803 (6627?)
num words 271005 3394686
xml file size 3.8MB 49MB
Retrieved from "https://wiki.apertium.org/w/index.php?title=Turkish_and_Kyrgyz/Kymorph_article&oldid=28839"