Difference between revisions of "Wolof/Transducer"

From LING073
Jump to: navigation, search
(create)
 
(→‎evaluation of morphological generation)
 
(21 intermediate revisions by 2 users not shown)
Line 1: Line 1:
 
Link to the code containing our transducer for Wolof: [https://github.swarthmore.edu/Ling073-sp23/ling073-wol github]
 
Link to the code containing our transducer for Wolof: [https://github.swarthmore.edu/Ling073-sp23/ling073-wol github]
−
======Evaluation======
 
  
−
======Notes======
+
======Notes and Evaluations======
 +
*Initial coverage of corpus: ~34%
 +
*By adding ci, di, ne, ak, Senegaal, and Yàlla, coverage went up to 42.3%.
 +
*The top unknown form is now "gu," with 70 occurrences. It's unclear to me exactly how I should analyze it, given that in the Peace Corps dictionary, it is defined as a relative pronoun (which, who, whom) on p15, and in the UCLA dictionary it is defined both as a relative subject that goes between g-class nouns and adjectives (that) and as an indefinite pronoun (any).
 +
*By fixing some errors in the code responsible for analyzing forms from our grammar page, coverage went up to 45.9%.
 +
 
 +
 
 +
*There are 161 stems in the transducer.
 +
*The list of most common unknown words is as follows:
 +
    TOP UNKNOWN WORDS:
 +
      70 ^gu/*gu$
 +
      67 ^ca/*ca$
 +
      57 ^mu/*mu$
 +
      49 ^juróom/*juróom$
 +
      46 ^ay/*ay$
 +
      39 ^at/*at$
 +
      38 ^te/*te$
 +
      38 ^nag/*nag$
 +
      37 ^suuf/*suuf$
 +
      35 ^Aji/*Aji$
 +
      32 ^réew/*réew$
 +
      31 ^Sax/*Sax$
 +
      30 ^nekk/*nekk$
 +
      29 ^am/*am$
 +
      25 ^téeméeri/*téeméeri$
 +
      25 ^a/*a$
 +
      24 ^jur/*jur$
 +
      22 ^Kayin/*Kayin$
 +
      20 ^muy/*muy$
 +
      20 ^loolu/*loolu$
 +
    coverage: 3367 / 7335 (~0.45903203817314246762)
 +
    remaining unknown forms: 3968
 +
 
 +
*Amount of tests matching gold for each test after finishing the analyzer:
 +
  Corpus 1 of 13: singular determiners-morph
 +
  16/16 (100.0%) tests pass (8/16 (50.0%) match gold)
 +
 
 +
  Corpus 2 of 13: plural determiners-morph
 +
  4/4 (100.0%) tests pass (2/4 (50.0%) match gold)
 +
 
 +
  Corpus 3 of 13: Personal Pronouns-morph
 +
  16/16 (100.0%) tests pass (0/16 (0.0%) match gold)
 +
 
 +
  Corpus 4 of 13: Demonstrative Pronouns-morph
 +
  3/3 (100.0%) tests pass (3/3 (100.0%) match gold)
 +
 
 +
  Corpus 5 of 13: Interrogrative Pronouns-morph
 +
  14/14 (100.0%) tests pass (14/14 (100.0%) match gold)
 +
 
 +
  Corpus 6 of 13: Indefinite Pronouns-morph
 +
  6/6 (100.0%) tests pass (6/6 (100.0%) match gold)
 +
 
 +
  Corpus 7 of 13: Possessive Pronouns-morph
 +
  11/11 (100.0%) tests pass (2/11 (18.18%) match gold)
 +
 
 +
  Corpus 8 of 13: Tense, Aspect, Mood marking-morph
 +
  22/22 (100.0%) tests pass (9/22 (40.91%) match gold)
 +
 
 +
  Corpus 9 of 13: Voices-morph
 +
  2/2 (100.0%) tests pass (0/2 (0.0%) match gold)
 +
 
 +
  Corpus 10 of 13: Proximity markers-morph
 +
  18/18 (100.0%) tests pass (6/18 (33.33%) match gold)
 +
 
 +
  Corpus 11 of 13: Subject focus-morph
 +
  4/4 (100.0%) tests pass (0/4 (0.0%) match gold)
 +
 
 +
  Corpus 12 of 13: Verb focus-morph
 +
  4/4 (100.0%) tests pass (0/4 (0.0%) match gold)
 +
 
 +
  Corpus 13 of 13: Object focus-morph
 +
  4/4 (100.0%) tests pass (2/4 (50.0%) match gold)
 +
 
 +
  All tests pass.
 +
 
 +
======evaluation of morphological generation======
 +
*passing morph and gen tests after finishing transducer, including one added twol rule
 +
*final coverage of corpus is 48.5344% after twol rule
 +
  Corpus 1 of 26: analysis
 +
  2/2 (100.0%) tests pass (0/2 (0.0%) match gold)
 +
 
 +
  Corpus 2 of 26: generation
 +
  1/1 (100.0%) tests pass (0/1 (0.0%) match gold)
 +
 
 +
  Corpus 3 of 26: singular determiners-morph
 +
  9/16 (56.25%) tests pass (8/9 (88.89%) match gold)
 +
 
 +
  Corpus 4 of 26: singular determiners-gen
 +
  16/16 (100.0%) tests pass (0/16 (0.0%) match gold)
 +
 
 +
  Corpus 5 of 26: plural determiners-morph
 +
  4/4 (100.0%) tests pass (2/4 (50.0%) match gold)
 +
 
 +
  Corpus 6 of 26: plural determiners-gen
 +
  4/4 (100.0%) tests pass (0/4 (0.0%) match gold)
 +
 
 +
  Corpus 7 of 26: Personal Pronouns-morph
 +
  14/16 (87.5%) tests pass (0/14 (0.0%) match gold)
 +
 
 +
  Corpus 8 of 26: Personal Pronouns-gen
 +
  18/18 (100.0%) tests pass (0/18 (0.0%) match gold)
 +
 
 +
  Corpus 9 of 26: Demonstrative Pronouns-morph
 +
  3/3 (100.0%) tests pass (3/3 (100.0%) match gold)
 +
 
 +
  Corpus 10 of 26: Demonstrative Pronouns-gen
 +
  3/3 (100.0%) tests pass (1/3 (33.33%) match gold)
 +
 
 +
  Corpus 11 of 26: Interrogrative Pronouns-morph
 +
  14/14 (100.0%) tests pass (14/14 (100.0%) match gold)
 +
 
 +
  Corpus 12 of 26: Interrogrative Pronouns-gen
 +
  14/14 (100.0%) tests pass (6/14 (42.86%) match gold)
 +
 
 +
  Corpus 13 of 26: Indefinite Pronouns-morph
 +
  6/6 (100.0%) tests pass (6/6 (100.0%) match gold)
 +
 
 +
  Corpus 14 of 26: Indefinite Pronouns-gen
 +
  6/6 (100.0%) tests pass (6/6 (100.0%) match gold)
 +
 
 +
  Corpus 15 of 26: Possessive Pronouns-morph
 +
  5/11 (45.45%) tests pass (4/5 (80.0%) match gold)
 +
 
 +
  Corpus 16 of 26: Possessive Pronouns-gen
 +
  5/15 (33.33%) tests pass (0/5 (0.0%) match gold)
 +
 
 +
  Corpus 17 of 26: Tense, Aspect, Mood marking-morph
 +
  21/22 (95.45%) tests pass (9/21 (42.86%) match gold)
 +
 
 +
  Corpus 18 of 26: Tense, Aspect, Mood marking-gen
 +
  27/27 (100.0%) tests pass (18/27 (66.67%) match gold)
 +
 
 +
  Corpus 19 of 26: Proximity markers-morph
 +
  8/18 (44.44%) tests pass (6/8 (75.0%) match gold)
 +
 
 +
  Corpus 20 of 26: Proximity markers-gen
 +
  20/20 (100.0%) tests pass (0/20 (0.0%) match gold)
 +
 
 +
  Corpus 21 of 26: Subject focus-morph
 +
  3/4 (75.0%) tests pass (0/3 (0.0%) match gold)
 +
 
 +
  Corpus 22 of 26: Subject focus-gen
 +
  4/4 (100.0%) tests pass (0/4 (0.0%) match gold)
 +
 
 +
  Corpus 23 of 26: Verb focus-morph
 +
  3/4 (75.0%) tests pass (0/3 (0.0%) match gold)
 +
 
 +
  Corpus 24 of 26: Verb focus-gen
 +
  4/4 (100.0%) tests pass (0/4 (0.0%) match gold)
 +
 
 +
  Corpus 25 of 26: Object focus-morph
 +
  2/3 (66.67%) tests pass (0/2 (0.0%) match gold)
 +
 
 +
  Corpus 26 of 26: Object focus-gen
 +
  3/3 (100.0%) tests pass (0/3 (0.0%) match gold)
 +
 +
 
 +
*Final evaluation of generator:
 +
 
 +
  Corpus 1 of 12: singular determiners-gen
 +
  16/16 (100.0%) tests pass (8/16 (50.0%) match gold)
 +
 
 +
  Corpus 2 of 12: plural determiners-gen
 +
  4/4 (100.0%) tests pass (2/4 (50.0%) match gold)
 +
 
 +
  Corpus 3 of 12: Personal Pronouns-gen
 +
  18/18 (100.0%) tests pass (0/18 (0.0%) match gold)
 +
 
 +
  Corpus 4 of 12: Demonstrative Pronouns-gen
 +
  3/3 (100.0%) tests pass (1/3 (33.33%) match gold)
 +
 
 +
  Corpus 5 of 12: Interrogrative Pronouns-gen
 +
  14/14 (100.0%) tests pass (6/14 (42.86%) match gold)
 +
 
 +
  Corpus 6 of 12: Indefinite Pronouns-gen
 +
  6/6 (100.0%) tests pass (6/6 (100.0%) match gold)
 +
 
 +
  Corpus 7 of 12: Possessive Pronouns-gen
 +
  5/15 (33.33%) tests pass (0/5 (0.0%) match gold)
 +
 
 +
  Corpus 8 of 12: Tense, Aspect, Mood marking-gen
 +
  27/27 (100.0%) tests pass (18/27 (66.67%) match gold)
 +
 
 +
  Corpus 9 of 12: Proximity markers-gen
 +
  20/20 (100.0%) tests pass (0/20 (0.0%) match gold)
 +
 
 +
  Corpus 10 of 12: Subject focus-gen
 +
  4/4 (100.0%) tests pass (0/4 (0.0%) match gold)
 +
 
 +
  Corpus 11 of 12: Verb focus-gen
 +
  4/4 (100.0%) tests pass (0/4 (0.0%) match gold)
 +
 
 +
  Corpus 12 of 12: Object focus-gen
 +
  3/3 (100.0%) tests pass (0/3 (0.0%) match gold)
 +
 
 +
 
 +
 
 +
[[Category:Sp23_Transducers]],[[Category:Wolof]]]

Latest revision as of 14:36, 6 April 2023

Link to the code containing our transducer for Wolof: github

Notes and Evaluations
  • Initial coverage of corpus: ~34%
  • By adding ci, di, ne, ak, Senegaal, and Yàlla, coverage went up to 42.3%.
  • The top unknown form is now "gu," with 70 occurrences. It's unclear to me exactly how I should analyze it, given that in the Peace Corps dictionary, it is defined as a relative pronoun (which, who, whom) on p15, and in the UCLA dictionary it is defined both as a relative subject that goes between g-class nouns and adjectives (that) and as an indefinite pronoun (any).
  • By fixing some errors in the code responsible for analyzing forms from our grammar page, coverage went up to 45.9%.


  • There are 161 stems in the transducer.
  • The list of most common unknown words is as follows:
    TOP UNKNOWN WORDS:
      70 ^gu/*gu$
      67 ^ca/*ca$
      57 ^mu/*mu$
      49 ^juróom/*juróom$
      46 ^ay/*ay$
      39 ^at/*at$
      38 ^te/*te$
      38 ^nag/*nag$
      37 ^suuf/*suuf$
      35 ^Aji/*Aji$
      32 ^réew/*réew$
      31 ^Sax/*Sax$
      30 ^nekk/*nekk$
      29 ^am/*am$
      25 ^téeméeri/*téeméeri$
      25 ^a/*a$
      24 ^jur/*jur$
      22 ^Kayin/*Kayin$
      20 ^muy/*muy$
      20 ^loolu/*loolu$
    coverage: 3367 / 7335 (~0.45903203817314246762)
    remaining unknown forms: 3968
  • Amount of tests matching gold for each test after finishing the analyzer:
 Corpus 1 of 13: singular determiners-morph
 16/16 (100.0%) tests pass (8/16 (50.0%) match gold)

 Corpus 2 of 13: plural determiners-morph
 4/4 (100.0%) tests pass (2/4 (50.0%) match gold)

 Corpus 3 of 13: Personal Pronouns-morph
 16/16 (100.0%) tests pass (0/16 (0.0%) match gold)

 Corpus 4 of 13: Demonstrative Pronouns-morph
 3/3 (100.0%) tests pass (3/3 (100.0%) match gold)

 Corpus 5 of 13: Interrogrative Pronouns-morph
 14/14 (100.0%) tests pass (14/14 (100.0%) match gold)

 Corpus 6 of 13: Indefinite Pronouns-morph
 6/6 (100.0%) tests pass (6/6 (100.0%) match gold)

 Corpus 7 of 13: Possessive Pronouns-morph
 11/11 (100.0%) tests pass (2/11 (18.18%) match gold)

 Corpus 8 of 13: Tense, Aspect, Mood marking-morph
 22/22 (100.0%) tests pass (9/22 (40.91%) match gold)

 Corpus 9 of 13: Voices-morph
 2/2 (100.0%) tests pass (0/2 (0.0%) match gold)

 Corpus 10 of 13: Proximity markers-morph
 18/18 (100.0%) tests pass (6/18 (33.33%) match gold)

 Corpus 11 of 13: Subject focus-morph
 4/4 (100.0%) tests pass (0/4 (0.0%) match gold)

 Corpus 12 of 13: Verb focus-morph
 4/4 (100.0%) tests pass (0/4 (0.0%) match gold)

 Corpus 13 of 13: Object focus-morph
 4/4 (100.0%) tests pass (2/4 (50.0%) match gold)

 All tests pass.
evaluation of morphological generation
  • passing morph and gen tests after finishing transducer, including one added twol rule
  • final coverage of corpus is 48.5344% after twol rule
 Corpus 1 of 26: analysis
 2/2 (100.0%) tests pass (0/2 (0.0%) match gold)

 Corpus 2 of 26: generation
 1/1 (100.0%) tests pass (0/1 (0.0%) match gold)

 Corpus 3 of 26: singular determiners-morph
 9/16 (56.25%) tests pass (8/9 (88.89%) match gold)

 Corpus 4 of 26: singular determiners-gen
 16/16 (100.0%) tests pass (0/16 (0.0%) match gold)

 Corpus 5 of 26: plural determiners-morph
 4/4 (100.0%) tests pass (2/4 (50.0%) match gold)

 Corpus 6 of 26: plural determiners-gen
 4/4 (100.0%) tests pass (0/4 (0.0%) match gold)

 Corpus 7 of 26: Personal Pronouns-morph
 14/16 (87.5%) tests pass (0/14 (0.0%) match gold)

 Corpus 8 of 26: Personal Pronouns-gen
 18/18 (100.0%) tests pass (0/18 (0.0%) match gold)

 Corpus 9 of 26: Demonstrative Pronouns-morph
 3/3 (100.0%) tests pass (3/3 (100.0%) match gold)

 Corpus 10 of 26: Demonstrative Pronouns-gen
 3/3 (100.0%) tests pass (1/3 (33.33%) match gold)

 Corpus 11 of 26: Interrogrative Pronouns-morph
 14/14 (100.0%) tests pass (14/14 (100.0%) match gold)

 Corpus 12 of 26: Interrogrative Pronouns-gen
 14/14 (100.0%) tests pass (6/14 (42.86%) match gold)

 Corpus 13 of 26: Indefinite Pronouns-morph
 6/6 (100.0%) tests pass (6/6 (100.0%) match gold)

 Corpus 14 of 26: Indefinite Pronouns-gen
 6/6 (100.0%) tests pass (6/6 (100.0%) match gold)

 Corpus 15 of 26: Possessive Pronouns-morph
 5/11 (45.45%) tests pass (4/5 (80.0%) match gold)

 Corpus 16 of 26: Possessive Pronouns-gen
 5/15 (33.33%) tests pass (0/5 (0.0%) match gold)

 Corpus 17 of 26: Tense, Aspect, Mood marking-morph
 21/22 (95.45%) tests pass (9/21 (42.86%) match gold)

 Corpus 18 of 26: Tense, Aspect, Mood marking-gen
 27/27 (100.0%) tests pass (18/27 (66.67%) match gold)

 Corpus 19 of 26: Proximity markers-morph
 8/18 (44.44%) tests pass (6/8 (75.0%) match gold)

 Corpus 20 of 26: Proximity markers-gen
 20/20 (100.0%) tests pass (0/20 (0.0%) match gold)

 Corpus 21 of 26: Subject focus-morph
 3/4 (75.0%) tests pass (0/3 (0.0%) match gold)

 Corpus 22 of 26: Subject focus-gen
 4/4 (100.0%) tests pass (0/4 (0.0%) match gold)

 Corpus 23 of 26: Verb focus-morph
 3/4 (75.0%) tests pass (0/3 (0.0%) match gold)

 Corpus 24 of 26: Verb focus-gen
 4/4 (100.0%) tests pass (0/4 (0.0%) match gold)

 Corpus 25 of 26: Object focus-morph
 2/3 (66.67%) tests pass (0/2 (0.0%) match gold)

 Corpus 26 of 26: Object focus-gen
 3/3 (100.0%) tests pass (0/3 (0.0%) match gold)
  • Final evaluation of generator:
 Corpus 1 of 12: singular determiners-gen
 16/16 (100.0%) tests pass (8/16 (50.0%) match gold)

 Corpus 2 of 12: plural determiners-gen
 4/4 (100.0%) tests pass (2/4 (50.0%) match gold)

 Corpus 3 of 12: Personal Pronouns-gen
 18/18 (100.0%) tests pass (0/18 (0.0%) match gold)

 Corpus 4 of 12: Demonstrative Pronouns-gen
 3/3 (100.0%) tests pass (1/3 (33.33%) match gold)

 Corpus 5 of 12: Interrogrative Pronouns-gen
 14/14 (100.0%) tests pass (6/14 (42.86%) match gold)

 Corpus 6 of 12: Indefinite Pronouns-gen
 6/6 (100.0%) tests pass (6/6 (100.0%) match gold)

 Corpus 7 of 12: Possessive Pronouns-gen
 5/15 (33.33%) tests pass (0/5 (0.0%) match gold)

 Corpus 8 of 12: Tense, Aspect, Mood marking-gen
 27/27 (100.0%) tests pass (18/27 (66.67%) match gold)

 Corpus 9 of 12: Proximity markers-gen
 20/20 (100.0%) tests pass (0/20 (0.0%) match gold)

 Corpus 10 of 12: Subject focus-gen
 4/4 (100.0%) tests pass (0/4 (0.0%) match gold)

 Corpus 11 of 12: Verb focus-gen
 4/4 (100.0%) tests pass (0/4 (0.0%) match gold)

 Corpus 12 of 12: Object focus-gen
 3/3 (100.0%) tests pass (0/3 (0.0%) match gold),]
Retrieved from "http://wikis.swarthmore.edu/ling073/index.php?title=Wolof/Transducer&oldid=20894"