Difference between revisions of "Turkish and Kyrgyz"
Jump to navigation
Jump to search
(14 intermediate revisions by 2 users not shown) | |||
Line 1: | Line 1: | ||
+ | {{TOCD}} |
||
− | ==Components== |
||
+ | |||
⚫ | |||
+ | ==Install== |
||
+ | |||
+ | You need: |
||
+ | |||
+ | * [[apertium-tur]] |
||
+ | * [[apertium-tur-kir]] |
||
+ | * [[apertium-kir]] |
||
+ | |||
+ | ===Usage=== |
||
+ | |||
+ | * [[Minimal installation from SVN]] (Apertium, lttoolbox) |
||
+ | * [[Apertium_and_Constraint_Grammar#Installing_VISL_CG3|Constraint Grammar]] |
||
⚫ | |||
+ | |||
+ | ==Transfer== |
||
+ | |||
+ | {|class=wikitable |
||
+ | ! t1x !! t2x !! t3x !! t4x !! t5x !! t6x |
||
+ | |- |
||
+ | |align=center| <code>transfer</code> ||align=center| <code>transfer</code> ||align=center| <code>chunker</code> ||align=center| <code>interchunk</code> ||align=center| <code>postchunk</code> ||align=center| <code>transfer</code> |
||
+ | |- |
||
+ | | collapse copula || compound tenses || chunking || verb valency || insert postpositions || joining clitics |
||
+ | |- |
||
+ | |} |
||
+ | |||
+ | ==Testing framework== |
||
+ | |||
+ | * Run <code>regression-tests.sh</code> and <code>pending-tests.sh</code> |
||
+ | * Run <code>sh qa.sh tur-kir</code>. |
||
+ | ** When you are happy with the result, copy <code>corpus/tur-kir-nova.txt</code> to <code>corpus/tur-kir-origina.txt</code> |
||
+ | ** Then <code>svn commit</code> |
||
+ | |||
+ | ==Todo== |
||
+ | |||
+ | * Properly review and deal with multiple derivations, e.g. <code>abil.pass</code>, <code>caus.pass</code> etc., especially where they require multiple modal verbs. |
||
+ | * [[Kyrgyz/Remaining_transducer_issues|Remaining Kyrgyz transducer issues]] |
||
==See also== |
==See also== |
||
Line 8: | Line 44: | ||
* [[/Regression tests|Regression tests]] |
* [[/Regression tests|Regression tests]] |
||
− | ;Some relevant HFST bugs |
||
− | |||
− | * [https://sourceforge.net/tracker/?func=detail&aid=3290585&group_id=224521&atid=1061990 hfst-proc does not have correct behaviour for joined words] |
||
− | * [https://sourceforge.net/tracker/?func=detail&aid=3269636&group_id=224521&atid=1061990 hfst-proc fails to analyse the cyrillic letter 'Э'] |
||
[[Category:Turkish and Kyrgyz|*]] |
[[Category:Turkish and Kyrgyz|*]] |
Latest revision as of 05:54, 20 April 2014
Install[edit]
You need:
Usage[edit]
- Minimal installation from SVN (Apertium, lttoolbox)
- Constraint Grammar
- HFST
Transfer[edit]
t1x | t2x | t3x | t4x | t5x | t6x |
---|---|---|---|---|---|
transfer |
transfer |
chunker |
interchunk |
postchunk |
transfer
|
collapse copula | compound tenses | chunking | verb valency | insert postpositions | joining clitics |
Testing framework[edit]
- Run
regression-tests.sh
andpending-tests.sh
- Run
sh qa.sh tur-kir
.- When you are happy with the result, copy
corpus/tur-kir-nova.txt
tocorpus/tur-kir-origina.txt
- Then
svn commit
- When you are happy with the result, copy
Todo[edit]
- Properly review and deal with multiple derivations, e.g.
abil.pass
,caus.pass
etc., especially where they require multiple modal verbs. - Remaining Kyrgyz transducer issues