GiellaLT provides an infrastructure for rule-based language technology aimed at minority and indigenous languages, and streamlines building anything from keyboards to speech technology. Read more about Why. See also our Privacy document.
This is an obsolete working plan from 2014, for historical reference.
This is an obsolete working plan from 2014, for historical reference.
Lexical selection
** Update existing sme-nob.lex to the new rule format: \ is done: apertium-sme-nob.sme-nob.lex
** Improve the rules: \ The old rules are moved into the new format. The work now is to make more rules
Lexical coverage
** Add dictionary and propernoun entries to the bidix
** Correct the entries:
_ add g3, nomag
_ correct the lemma form to be the same as in lexc
*** add remaining lemmas to lexc
** Start looking at lexicon holes
Linguistic rules
** is next
Semantic tags, we want
** semtags for the disambiguation and the lexical selection rules
** semtags after the PoS
** only one semtag for each N, that means that lemmas with more than one, e.g. Sem/Org and Sem/Plc, must get a common one, e.g. Sem/OrgPlc
** a semtag for all nouns, that means a dummy for those which don’t have any yet: Sem/Dummy ?
** We have to add semtags to the bidix
** Alternative would be generating sets from LEXC, according to the semtags. These sets have to be added both to the disambiguation+functions-file and to the apertium-sme-nob.sme-nob.lex