User:Capsot/GSOC 2018 Data
Jump to navigation
Jump to search
Statistics
| Bidix | Coverage | WER | |
| fra-oci | 41,000 (27,000 without family names) | 92.3% | 10% |
| oci-fra | 41,000 (27,000 without family names | 92.9% | Not calculated (due to non-functional morphological disambiguator) |
Rules
| Disambiguation Rules | Lexical Selection Rules | Transfer Rules | |
| fra-oci | 678 | 855 (85 without counting the selection rules for anthroponyms) | t1x: 130, t2x: 14, t3x: 1, t4x: 5 (for punctuation) |
| oci-fra | 87 | 928 (151 without counting the selection rules for anthroponyms) | t1x: 51, t2ax: 21 (mostly for the insertion or not of the subject pronoun and agreement between subject and attribute), t2bx: 4 (inclusion or not of adverb "ne" in negation), t2cx: 1 (for the partitive article after verb), t3x: 2, t4x: 5 (for punctuation) |
- PS: (oci-fra) A corpus of 14,000 words has been manually disambiguated for getting a morphological disambuagator, but we couldn't get a working prob file.
