Kalmyk/Transducer
From LING073
Contents
Code
https://github.swarthmore.edu/slim1/ling073-xal
Analyzer Evaluation
Evaluation
The total number of stems in the transducer is 56.
The current coverage is 5.62% of 1138 tokenized words.
The top unknown words are:
| Word | Occurrences |
|---|---|
| Бурхн | 16 |
| Хальмг | 14 |
| ик | 12 |
| Эн | 12 |
| Зая | 12 |
| болтха | 12 |
| төлә | 10 |
| болв | 10 |
| н | 8 |
| Би | 7 |
| андит | 7 |
| күн | 7 |
| тер | 6 |
| һазр | 6 |
| уга | 6 |
| анав | 6 |
| сән | 6 |
| бәә | 6 |
| Деед | 6 |
| келв | 6 |
All tests in the main yaml file pass.
1 test in the commonwords.yaml file passes.
Notes
All tests currently pass.
Words Added
- By adding болн<conjcoo>: болн ("and") to the transducer, coverage rate went from 4.57% to 5.62%.
Generator Evaluation
Initial Evaluation of Morphological Generation
Morphological Analysis Test
- Passes: 122
- Fails: 3
- Total: 125
Corpus Coverage Test
- Corpus Coverage: 6.94%
Morphological Generation Test
- Passes: 122
- Fails: 69
- Total: 191
Final Evaluation of Morphological Generation
Morphological Generation Test
- Passes: 107
- Fails: 29
- Total: 136
- Number of twol rules added: 9
Corpus Coverage Test
- Corpus Coverage: 5.18%
