Occitan and French/Work plan

From Apertium
Revision as of 07:13, 16 March 2018 by Hectoralos (talk | contribs)
Jump to navigation Jump to search

Note: The French → Occitan part of the project is the main direction.

Setmana Dates Descripció Bidix
(sense np)
previst
(%) Cobertura
prevista
(%) WER
previst
Testvoc Avaluació Bidix
real
(%) Cobertura
real
(%) WER Err. Fet?
0 français > occitan ~5 700
1 14 mai—20 mai ~10 000 ~84,0%
2 21 mai—11 mai ~12 000 ~86,0%
3 28 mai—3 junh ~14 000 ~88.0%
4 4 junh—10 junh ~16 000 ~89.0%
5 11 junh—15 junh ~18 000 ~89.5% ~35% pr, cnj*, adv
6 18 junh—24 junh ~20 000 ~90.0% prn, det
7 25 junh—1 julhet ~21 000 ~90.5% vblex
8 2 julhet—8 julhet ~22 000 ~91.0% adj
9 9 julhet—13 julhet ~22 000 ~91.0% ~25% n
0 occitan > français ~20 000
10 16 julhet—22 julhet ~20 500 ~88.0% pr, cnj*, adv
13 23 julhet—29 julhet ~21 000 ~89.0% prn, det, n, adj
11 30 julhet—5 agost ~21 500 ~90.0% vblex
12* 6 agost—9 agost ~21.500 ~90.5%
12** 10 agost—14 agost avaluacion finala ~35%

Per a calcular els nombres

Errors (calcular en apertium-fra-oci/dev)
$ bash dev/testvoc/generation.sh fra-oci | wc -l  # en apertium-oci-fra
$ bash dev/testvoc/generation.sh oci-fra | wc -l  # en apertium-oci-fra
Bidix (calcular en apertium-oci-fra)
$ cat apertium-oci-fra.oci-fra.dix | grep '<l'  | grep -v '¨np"' | wc -l
Cobertura (calcular en apertium-oci-fra)
$ cat ../apertium-fra/corpus/corpus_fra_wp100000.txt | apertium -d . fra-oci-morph | sed 's/\$\W*\^/$\n^/g' > /tmp/fra-oci.coverage.txt
$ calc `cat /tmp/fra-oci.coverage.txt | grep -v '\*' | wc -l `/`cat /tmp/fra-oci.coverage.txt | wc -l`

$ cat ../apertium-cat/corpus/corpus_oci_wp100000.txt | apertium -d . oci-fra-morph | sed 's/\$\W*\^/$\n^/g' > /tmp/oci-fra.coverage.txt
$ calc `cat /tmp/oci-fra.coverage.txt | grep -v '\*' | wc -l `/`cat /tmp/oci-fra.coverage.txt | wc -l`

Veire tanben