Huaylas Quechua/Transducer

link to code

Analyzer Evaluation

On 3/17, coverage is 10.06%

Number of stems:

• Lexicons: 19
• Lexicon entries: 93
• Patterns: 2
• Pattern entries: 7

Number of initial evaluation of morphological generationtests that pass:

• Plural-morph: 0/3 (0.0%) tests pass

• Possession-morph: 9/12 (75.0%) tests pass (9/9 (100.0%) match gold)

• Combining Inflections-morph: 3/3 (100.0%) tests pass (3/3 (100.0%) match gold)

• Present Tense-morph: 5/21 (23.81%) tests pass (3/5 (60.0%) match gold)

• Preterite Tense-morph: 11/19 (57.89%) tests pass (3/11 (27.27%) match gold)

• Future Tense-morph: 9/20 (45.0%) tests pass (2/9 (22.22%) match gold)

• Present Tense Transitions-morph: 0/15 (0.0%) tests pass

• Past Tense Transitions-morph: 12/15 (80.0%) tests pass (0/12 (0.0%) match gold)

• Future Tense Transitions-morph: 3/15 (20.0%) tests pass (0/3 (0.0%) match gold)

• Case-morph: 0/12 (0.0%) tests pass

Note: The Case-morph and Plural-morph tests are not passing despite the correct forms being in the transducer matching the forms in the gold.txt file.

Top unknown words in the corpus:

• 34 Teyta
• 32 Dios
• 31 y
• 14 nirqan
• 14 Tsëpitanam
• 11 alli
• 9 kë
• 9 Tsë
• 9 tukïläya
• 8 mana
• 8 llapanpis
• 8 wätana
• 7 Tsëmi
• 7 Tsëkunatam
• 7 junaqchö
• 7 Ninqannöllam
• 6 kamarirmi
• 6 kanqanta
• 6 këkan
• 5 yaku

Generator Evaluation

initial evaluation of morphological generation

See Analyzer Evaluation for stats on the tests that passed

final evaluation of morphological generation

• coverage of corpus: 10.06%
• number of tests passing:

• number of twol rules added:

• precision: 100%
• recall: 50.76%