Difference between revisions of "Adyghe/Transducer"

From LING073
Jump to: navigation, search
(Top Unknown Forms)
(Notes)
 
(20 intermediate revisions by the same user not shown)
Line 1: Line 1:
 
==Analyzer Evaluation==
 
==Analyzer Evaluation==
 
[https://github.swarthmore.edu/Ling073-sp22/ling073-ady Our Transducer]
 
[https://github.swarthmore.edu/Ling073-sp22/ling073-ady Our Transducer]
 +
====Status====
 +
{| class="wikitable" style="text-align: center
 +
|'''Date'''
 +
|'''Patterns'''
 +
|'''Nouns'''
 +
|'''Verbs'''
 +
|'''Adjectives'''
 +
|'''Pronouns'''
 +
|'''Demonstratives'''
 +
|'''Twol Rules'''
 +
|'''Basic Coverage'''
 +
|'''Bible Coverage'''
 +
|'''Tests'''
 +
|-
 +
|'' May 17 12:53 EDT ''
 +
|''12''
 +
|''34''
 +
|''21''
 +
|''4''
 +
|''4''
 +
|''3''
 +
|''12''
 +
|''134/779''
 +
|''143/449''
 +
|''44/60''
 +
|-
 +
|'' May 17 14:26 EDT ''
 +
|''12''
 +
|''34''
 +
|''21''
 +
|''4''
 +
|''4''
 +
|''3''
 +
|''13''
 +
|''134/779''
 +
|''143/449''
 +
|''44/60''
 +
|}
 +
 
====Basic Corpus====
 
====Basic Corpus====
 
{| class="wikitable" style="text-align: center
 
{| class="wikitable" style="text-align: center
Line 7: Line 46:
 
|'''Remaining unknown forms'''
 
|'''Remaining unknown forms'''
 
|-
 
|-
|'' May 17 12:24 ''
+
|'' May 17 12:24 EDT ''
 
|''134/779 (~0.17201540436)''
 
|''134/779 (~0.17201540436)''
 
|''645''
 
|''645''
Line 13: Line 52:
  
 
====Bible Corpus====
 
====Bible Corpus====
Coverage: 134/449
+
{| class="wikitable" style="text-align: center
 +
|'''Date Evaluated'''
 +
|'''Coverage'''
 +
|'''Remaining unknown forms'''
 +
|-
 +
|'' May 17 12:46 EDT ''
 +
|''143/449 (~0.3184855233853)''
 +
|''306''
 +
|}
  
 
====Top Unknown Forms====
 
====Top Unknown Forms====
Line 39: Line 86:
  
 
Bible Corpus
 
Bible Corpus
 +
    13 ^къэхъугъ/*къэхъугъ$
 +
      9 ^ошъо/*ошъо$
 +
      6 ^хъугъэ/*хъугъэ$
 +
      6 ^пчэдыжь/*пчэдыжь$
 +
      6 ^лъэпкъ/*лъэпкъ$
 +
      6 ^кӏэ/*кӏэ$
 +
      6 ^Аущтэуи/*Аущтэуи$
 +
      5 ^къыгъэшӏыгъ/*къыгъэшӏыгъ$
 +
      5 ^зэкӏэ/*зэкӏэ$
 +
      5 ^зиӏэ/*зиӏэ$
 +
      4 ^псэу/*псэу$
 +
      4 ^еджагъ/*еджагъ$
 +
      3 ^щыцохъорэ/*щыцохъорэ$
 +
      3 ^щыпшырэ/*щыпшырэ$
 +
      3 ^фэдэрэ/*фэдэрэ$
 +
      3 ^сыд/*сыд$
 +
      3 ^къыгъэшӏыгъэх/*къыгъэшӏыгъэх$
 +
      3 ^зэфэшъхьафхэри/*зэфэшъхьафхэри$
 +
      3 ^зэфэшъхьафхэр/*зэфэшъхьафхэр$
 +
      2 ^ычӏэгъ/*ычӏэгъ$
  
 
==MorphTest Evaluation==
 
==MorphTest Evaluation==
Line 59: Line 126:
 
|44
 
|44
 
|18
 
|18
 +
|62
 +
|70.9%
 +
|-
 +
|rowspan="2"|May 17 2022 14:26
 +
|''Analyser''
 +
|44
 +
|16
 
|60
 
|60
|70.9%
+
|73.3%
 +
|-
 +
|''Generator''
 +
|44
 +
|16
 +
|60
 +
|73.3%
 
|}
 
|}
  
 
==Notes==
 
==Notes==
Overall - We have worked mainly on nouns, as the verbal system is highly morphologically and phonologically complex (and the only extensive source on verbal morphology is written in Russian).
+
Overall - We have worked mainly on nouns, as the verbal system is highly morphologically and phonologically complex (and the only extensive source on verbal morphology is written in Russian). Unfortunately, the MorphTests may no longer pass, as we have changed the order in which we output noun tags to accommodate postpositions.
  
 
Corpus Coverage -  
 
Corpus Coverage -  
  
 
MorphTests - Second-person prefixes in Adyghe always have multiple forms based on the phonological environment. The generator is currently producing both forms for each possessed noun.
 
MorphTests - Second-person prefixes in Adyghe always have multiple forms based on the phonological environment. The generator is currently producing both forms for each possessed noun.
 +
 +
Rules - Unfortunately, we did not document the change in our corpus coverage as we added twol rules. Three of our twol rules were added to increase the number of MorphTest passes, and the rest are used in basic verbal morphology.
  
  
 
[[Category:Adyghe]][[Category:Sp22 Transducers]]
 
[[Category:Adyghe]][[Category:Sp22 Transducers]]

Latest revision as of 10:21, 18 May 2022

Analyzer Evaluation

Our Transducer

Status

Date Patterns Nouns Verbs Adjectives Pronouns Demonstratives Twol Rules Basic Coverage Bible Coverage Tests
May 17 12:53 EDT 12 34 21 4 4 3 12 134/779 143/449 44/60
May 17 14:26 EDT 12 34 21 4 4 3 13 134/779 143/449 44/60

Basic Corpus

Date Evaluated Coverage Remaining unknown forms
May 17 12:24 EDT 134/779 (~0.17201540436) 645

Bible Corpus

Date Evaluated Coverage Remaining unknown forms
May 17 12:46 EDT 143/449 (~0.3184855233853) 306

Top Unknown Forms

Basic Corpus

    17 ^е/*е$
    15 ^цIыфи/*цIыфи$
    15 ^Хэтрэ/*Хэтрэ$
    13 ^хъущтэп/*хъущтэп$
    12 ^иІ/*иІ$
     9 ^къэхъугъ/*къэхъугъ$
     9 ^зы/*зы$
     9 ^Хэти/*Хэти$
     8 ^ошъо/*ошъо$
     6 ^къэралым/*къэралым$
     5 ^зыми/*зыми$
     4 ^ыкІи/*ыкІи$
     4 ^хэтрэ/*хэтрэ$
     4 ^хъугъэ/*хъугъэ$
     4 ^фэдэ/*фэдэ$
     4 ^сыд/*сыд$
     4 ^пчэдыжь/*пчэдыжь$
     4 ^псэу/*псэу$
     4 ^лъэпкъ/*лъэпкъ$
     4 ^кӏэ/*кӏэ$

Bible Corpus

    13 ^къэхъугъ/*къэхъугъ$
     9 ^ошъо/*ошъо$
     6 ^хъугъэ/*хъугъэ$
     6 ^пчэдыжь/*пчэдыжь$
     6 ^лъэпкъ/*лъэпкъ$
     6 ^кӏэ/*кӏэ$
     6 ^Аущтэуи/*Аущтэуи$
     5 ^къыгъэшӏыгъ/*къыгъэшӏыгъ$
     5 ^зэкӏэ/*зэкӏэ$
     5 ^зиӏэ/*зиӏэ$
     4 ^псэу/*псэу$
     4 ^еджагъ/*еджагъ$
     3 ^щыцохъорэ/*щыцохъорэ$
     3 ^щыпшырэ/*щыпшырэ$
     3 ^фэдэрэ/*фэдэрэ$
     3 ^сыд/*сыд$
     3 ^къыгъэшӏыгъэх/*къыгъэшӏыгъэх$
     3 ^зэфэшъхьафхэри/*зэфэшъхьафхэри$
     3 ^зэфэшъхьафхэр/*зэфэшъхьафхэр$
     2 ^ычӏэгъ/*ычӏэгъ$

MorphTest Evaluation

Date Evaluated Evaluation Total passes Total fails Total Percentage
May 17 2022 12:30 Analyser 44 16 60 73.3%
Generator 44 18 62 70.9%
May 17 2022 14:26 Analyser 44 16 60 73.3%
Generator 44 16 60 73.3%

Notes

Overall - We have worked mainly on nouns, as the verbal system is highly morphologically and phonologically complex (and the only extensive source on verbal morphology is written in Russian). Unfortunately, the MorphTests may no longer pass, as we have changed the order in which we output noun tags to accommodate postpositions.

Corpus Coverage -

MorphTests - Second-person prefixes in Adyghe always have multiple forms based on the phonological environment. The generator is currently producing both forms for each possessed noun.

Rules - Unfortunately, we did not document the change in our corpus coverage as we added twol rules. Three of our twol rules were added to increase the number of MorphTest passes, and the rest are used in basic verbal morphology.