Difference between revisions of "RFERL corpora"

From Apertium
Jump to navigation Jump to search
 
(4 intermediate revisions by the same user not shown)
Line 1: Line 1:
Radio Free Europe / Radio Liberty runs news services in a number of Central Asian languages. The information is essentially free for public use with attribution. {{comment|link to usage info}}
Radio Free Europe / Radio Liberty runs news services in a number of Central Asian languages. The information is essentially free for public use with attribution. {{comment|need link to usage info}}

We discovered [[Turkish and Kyrgyz/Making a corpus from azattyk|how a corpus could be built from their website]], and have instructions for [[writing a scraper]] in the framework we developed for it. Currently we have corpora for Kazakh and Kyrgyz, covering only a couple years' worth of articles.


== Kyrgyz ==
== Kyrgyz ==
* Site: [http://azattyk.org azattyk.org]
* Site: [http://azattyk.org azattyk.org]
* Coverage with: [[kymorph]]


=== 2009 ===
=== 2009 ===
* Number of stems: {{:RFERL corpus/ky/2009/stems}}
* Number of stems: {{:RFERL corpus/ky/2009/stems}}
* Coverage: {{:Kymorph/coverage/rferl2009}}
* Coverage: ~{{:Kymorph/coverage/rferl2009}}%


=== 2010 ===
=== 2010 ===
* Number of stems: {{:RFERL corpus/ky/2010/stems}}
* Number of stems: {{:RFERL corpus/ky/2010/stems}}
* Coverage: {{:Kymorph/coverage/rferl2010}}
* Coverage: ~{{:Kymorph/coverage/rferl2010}}%


== Kazakh ==
== Kazakh ==
Line 17: Line 20:
=== 2009 ===
=== 2009 ===
* Number of stems: {{:RFERL corpus/kk/2009/stems}}
* Number of stems: {{:RFERL corpus/kk/2009/stems}}
* Coverage: {{:Kazmorph/coverage/rferl2009}}
* Coverage: ~{{:Kazmorph/coverage/rferl2009}}%


=== 2010 ===
=== 2010 ===
* Number of stems: {{:RFERL corpus/kk/2010/stems}}
* Number of stems: {{:RFERL corpus/kk/2010/stems}}
* Coverage: {{:Kazmorph/coverage/rferl2010}}
* Coverage: ~{{:Kazmorph/coverage/rferl2010}}%

Latest revision as of 18:41, 5 December 2013

Radio Free Europe / Radio Liberty runs news services in a number of Central Asian languages. The information is essentially free for public use with attribution.

need link to usage info

We discovered how a corpus could be built from their website, and have instructions for writing a scraper in the framework we developed for it. Currently we have corpora for Kazakh and Kyrgyz, covering only a couple years' worth of articles.

Kyrgyz[edit]

2009[edit]

  • Number of stems: 4.1M
  • Coverage: ~87.4%

2010[edit]

  • Number of stems: 3.4M
  • Coverage: ~88%

Kazakh[edit]

2009[edit]

2010[edit]

  • Number of stems: 3.2M
  • Coverage: ~85.4%