Difference between revisions of "Walloon/Transducer"
From LING073
| (13 intermediate revisions by the same user not shown) | |||
| Line 2: | Line 2: | ||
* [https://github.swarthmore.edu/Ling073-sp25/ling073-wln GitHub Repository] | * [https://github.swarthmore.edu/Ling073-sp25/ling073-wln GitHub Repository] | ||
| − | + | * Note: <code>hfst-fst2txt -f dot wln.autogen.hfst | dot -Tpdf -o graph.pdf</code> | |
| − | == | + | == Analyser Evaluation == |
| + | Number of Stems: 42 | ||
| + | |||
| + | Current Coverage Over Corpus: 114 / 1357 | ||
| + | |||
| + | Current List of Top Unknown Words: | ||
| + | * l<unk> | ||
| + | * di<unk> | ||
| + | * k<unk> | ||
| + | * ki<unk> | ||
| + | * et<unk> | ||
| + | * d<unk> | ||
| + | * a<unk> | ||
| + | * i<unk> | ||
| + | * s<unk> | ||
| + | * li<unk> | ||
| + | * po<unk> | ||
| + | * n<unk> | ||
| + | * tot<unk> | ||
| + | * ont<unk> | ||
| + | * å<unk> | ||
| + | * est<unk> | ||
| + | * vs<unk> | ||
| + | * nén<unk> | ||
| + | * I<unk> | ||
| + | * vos<unk> | ||
| + | |||
| + | Number of Tests Passing: ~35% (some tests are failing to register as passing despite correct form being outputted via <code>hfst-expand -c0 wln.automorf.hfst</code>) | ||
| + | |||
| + | Note: Could not add for "Regularly Irregular Verbs" due to complexity (requires counting amount of consonants) | ||
| + | |||
| + | == Generator Evaluation == | ||
| + | === Initial Evaluation of Morphological Generation === | ||
| + | Started work before tagging "analyser" by mistake, and so unsure of exact number passing. Was significantly lower than final, though. | ||
| + | |||
| + | === Final Evaluation of Morphological Generation === | ||
| + | * Passing Tests: 54/54 (100%) | ||
| + | * Tests Matching Gold: 47/54 (~87%) | ||
| + | * Number of Twol rules added: 4 | ||
| + | * Coverage Over Corpus: 114 / 1357 (unchanged) | ||
[[Category:Sp25 Transducers]] [[Category: Walloon]] | [[Category:Sp25 Transducers]] [[Category: Walloon]] | ||
Latest revision as of 03:32, 20 March 2025
Contents
Overview
- Note:
hfst-fst2txt -f dot wln.autogen.hfst | dot -Tpdf -o graph.pdf
Analyser Evaluation
Number of Stems: 42
Current Coverage Over Corpus: 114 / 1357
Current List of Top Unknown Words:
- l<unk>
- di<unk>
- k<unk>
- ki<unk>
- et<unk>
- d<unk>
- a<unk>
- i<unk>
- s<unk>
- li<unk>
- po<unk>
- n<unk>
- tot<unk>
- ont<unk>
- å<unk>
- est<unk>
- vs<unk>
- nén<unk>
- I<unk>
- vos<unk>
Number of Tests Passing: ~35% (some tests are failing to register as passing despite correct form being outputted via hfst-expand -c0 wln.automorf.hfst)
Note: Could not add for "Regularly Irregular Verbs" due to complexity (requires counting amount of consonants)
Generator Evaluation
Initial Evaluation of Morphological Generation
Started work before tagging "analyser" by mistake, and so unsure of exact number passing. Was significantly lower than final, though.
Final Evaluation of Morphological Generation
- Passing Tests: 54/54 (100%)
- Tests Matching Gold: 47/54 (~87%)
- Number of Twol rules added: 4
- Coverage Over Corpus: 114 / 1357 (unchanged)
