Language Technology at UiT The Arctic University of Norway

The Divvun and Giellatekno teams build language technology aimed at minority and indigenous languages

View GiellaLT on GitHub divvungiellatekno/giellalt.uit.no

Page Content

Grammatikkontrollmøte 5.9.2018

Til stades: Duommá, Kevin, Linda, Sjur

Saker:

Status

Duommá

Linda

Følgende formen blir ikke generert (sjå punkt lenger ned):

"<jierbmát>"
        "jierbmái" Ex/A Sem/Hum Der/AAdv Adv Err/Orth <W:0.0000000000> <ctjHead> <ctjHead> @ADVL> SELECT:11627:r1241 MAP:22182 &typo #3->3 ADD:3822:Err/Orth-any SUBSTITUTE:8118:ctjHead SUBSTITUTE:8118:ctjHead
typo
        "jierbmái" Sem/Hum Der/AAdv Adv <W:0.0000000000> <ctjHead> <ctjHead> @ADVL> SELECT:11627:r1241 MAP:22182 &typo &SUGGEST #3->3 ADD:3822:Err/Orth-any SUBSTITUTE:8118:ctjHead SUBSTITUTE:8118:ctjHead COPY:13036:Err/Orth
jierbmái+Der/AAdv+Adv   ?

Så problemet er at


manglar frå `tools/grammarcheckers/generator-gramcheck-gt-norm.hfstol` – det
ser ut som han er med i ein del av generatorane i langs/sme, men ikkje alle:

== src/generator-gramcheck-gt-norm.hfstol: == jierbmái+A+Der/AAdv+Adv jierbmái+A+Der/AAdv+Adv+? inf


Det skal vera: `jierbmái+Ex/A+Der/AAdv+Adv`. Og då´funkar det.

Analysetaggen har vorte endra av ein regel, kanskje denne:

LIST Err/Orth-any = Err/Orth Err/Orth-a-á Err/Orth-nom-gen Err/Orth-nom-acc ; COPY (A &SUGGEST) EXCEPT (Err/Orth Ex/A) AFTER (“.“r) TARGET Err/Orth-any + (&typo) + Ex/A; →burde vel vera: COPY (Adv &SUGGEST) EXCEPT (Err/Orth) AFTER (“.“r) TARGET Err/Orth-any + (&typo) + Adv;


Gammal regel, fjerna 19.juli:

-COPY:Err/Orth-any (&SUGGEST) EXCEPT Err/Orth-any_OR_Ex TARGET Err/Orth-any + (&typo) ; Ny: +COPY:Err/Orth-any (N &SUGGEST) EXCEPT (Err/Orth Ex/N) OR (Err/Orth-a-á Ex/N) OR (Err/Orth-nom-gen Ex/N) OR (Err/Orth-nom-acc Ex/N) AFTER (“.*“r) TARGET Err/Orth-any + (&typo) + Ex/N ; osb.


Setninga som illustrerer problemet:

Filmmat leat jierbmát ja bures válljejuvvon, go dat leat filmmat mat leat dahkan sámiid eanemus dovddus, dadjá Store Jakobsen.


##  Kevin

* WoodWing vs soft hyphens (bør fungera for Ávvir no; dei spør om støtte for digitaleditor òg)

##  Sjur

* Har forbetra orddelings-fst-ane vi genererer (og funne feil -> Duommá)
* URL-analysane får no riktig CG-tagg i analysen frå hfst-tokeniser

#  Orddeling

Vi utsett arbeidet med å leggja til orddeling. Det hastar ikkje no.

#  r168875, r168876

Innsjekkingsmelding:

*«I had to move these rules from the spellchecker.cg3 grammar to grammarchecker.cg3. The reason is that once a file is mapping or adding tags with the same prefix, the next grammar cannot add further tags with the same prefix. At least that is my conclusion after testing various options. So once typo was mapped, I could not add an errortag for a superlative error anymore. In another pipeline without spellchecker.cg3 it worked fine though. Maybe this creates problems if we use the pipeline without grammarchecker.cg3 so we might need to talk about this at some point.»*

Relevante reglar:

ADD:spelled (&typo &SUGGESTWF) () ; ADD:Err/Orth-any (&typo) Err/Orth-any (NOT 0 (*) - Err/Orth-any) ; COPY:Err/Orth-any (N &SUGGEST) EXCEPT (Err/Orth Ex/N) OR (Err/Orth-a-á Ex/N) OR (Err/Orth-nom-gen Ex/N) OR (Err/Orth-nom-acc Ex/N) AFTER (".*"r) TARGET Err/Orth-any + (&typo) + Ex/N ; COPY:Err/Orth-any (V &SUGGEST) EXCEPT (Err/Orth Ex/V) OR (Err/Orth-a-á Ex/V) OR (Err/Orth-nom-gen Ex/V) OR (Err/Orth-nom-acc Ex/V) AFTER (".*"r) TARGET Err/Orth-any + (&typo) + Ex/V ; COPY:Err/Orth-any (A &SUGGEST) EXCEPT (Err/Orth Ex/A) OR (Err/Orth-a-á Ex/A) OR (Err/Orth-nom-gen Ex/A) OR (Err/Orth-nom-acc Ex/A) AFTER (".*"r) TARGET Err/Orth-any + (&typo) + Ex/A; COPY:Err/Orth-any (IV &SUGGEST) EXCEPT (Err/Orth Ex/IV) OR (Err/Orth-a-á Ex/IV) OR (Err/Orth-nom-gen Ex/IV) OR (Err/Orth-nom-acc Ex/IV) AFTER (".*"r) TARGET Err/Orth-any + (&typo) + Ex/IV; COPY:Err/Orth-any (TV &SUGGEST) EXCEPT (Err/Orth Ex/TV) OR (Err/Orth-a-á Ex/TV) OR (Err/Orth-nom-gen Ex/TV) OR (Err/Orth-nom-acc Ex/TV) AFTER (".*"r) TARGET Err/Orth-any + (&typo) + Ex/TV; ```

Arbeid framover

Duommá

Linda

Kevin

Eckhard til Tromsø

Vi tek opp det seinare.