Difference between revisions of "Guarani and Warlpiri"
From LING073
(→gug → wbp Evaluation) |
(→Results when running on sentences (corpus)) |
||
| Line 22: | Line 22: | ||
*PER = 100% | *PER = 100% | ||
===Results when running on sentences (corpus) === | ===Results when running on sentences (corpus) === | ||
| + | *Number of words in reference: 58 | ||
When removing unknown words: | When removing unknown words: | ||
*WER = 22.41% | *WER = 22.41% | ||
Revision as of 00:43, 26 March 2017
Resources for machine translation between Guarani and Warlpiri
Contents
Corpora
wbp -> gug evaluation
Results when running on tests
- Number of unknown words: 4
- Percentage of unknown words: 28.57%
When removing unknown words:
- WER = 78.57%
- PER = 78.57%
When unknown words are not removed:
- WER = 100%
- PER = 100%
Results when running on sentences (corpus)
- Number of words in reference: 58
When removing unknown words:
- WER = 22.41%
- PER = 22.41%
When unknown words are not removed:
- WER = 100%
- PER = 100%
Developed Resources
- Corpus Repository
- Contrastive Grammar
- Guarani to Warlpiri analyzer repository
- Warlpiri to Guarani analyzer repository
Lexical Selection
gug → wbp Evaluation
Results when running on tests
- Number of unknown words: 0
- Percentage of unknown words: 0%
When removing unknown words:
- WER = 92.86%
- PER = 92.86%
When unknown words are not removed:
- WER = 92.86%
- PER = 92.86%
Results when running on sentences (corpus)
- Number of unknown words: 120
- Percentage of unknown words: 206.90%
When removing unknown words:
- WER = 0.0%
- PER = 0.0%
When unknown words are not removed:
- WER = 100%
- PER = 100%