Faroese disambiguator

On this page

• Delimiters, tags and sets
• Delimiters
• Tags and sets
  • Tags
  • Sets
    • Noun sets
    • Adjective sets
    • Nominal sets
    • Verb sets
    • Noun-Verb sets
    • Number sets
    • Preposition sets, taking different cases
  • Boundary sets
  • Case sets
  • Word sets
  • Early and popular rules
  • Infinitive
  • A or N
  • Conjunctions
  • Subjunctions
  • Verbs
    • Passive
• PP disambiguation
• Case within PP phrases
• POS disambiguation
  • Pron not N
  • Proper nouns
• Specific lexemes, words
• Adverbs
  • General adverb
  • Specific adverbs
  • Lexicalised adverbs.
  • Adverb verbs
  • Idioms
• Numerals and number symbols
• NP internal constraints
  • Determiner disambiguation
  • Specific determiners
    • Postnominal determiner disambiguation
  • Definiteness disambiguation
• Case disambiguation
  • Noun disambiguation
  • Poss disambiguation
  • Number disambiguation
  • Coordination
  • Embedded clause V topicalisation
  • Elliptic AP as NP
  • P chains or not
  • Pronoun disambiguation
  • NP Coordination
• VP disambiguation
  • V or A
  • V or Adv
  • V or N
  • Infinitive
  • Imperative
  • Present participle
  • Supine
  • Present singular
  • Present plural
  • V + Refl
  • Past indicative
• Case disambiguation
  • Nominative
    • Predicative
    • Subject
    • Miscellanious
  • Accusative
    • Mær dámar
  • Genitive
• Pronoun disambiguation
• Verb disambiguation
  • A or V
  • Person
  • Number disambiguation
  • Postverbal subject
  • Gender disamb of adjectives
  • Gender disamb of numerals
  • Case disamb of numerals
  • Ordinals
  • Coordination
  • Adjective disambiguation outside NP
• Substituting tags
  • Titles
  • CC Coordinate NPs
• For Apertium

Faroese disambiguator

Usage, in lang-fao: cat text.txt|hfst-tokenize -cg tools/tokenisers/tokeniser-disamb-gt-desc.pmhfst |vislcg3 -g src/cg3/disambiguator.cg3

This file documents the Faroese disambiguator file .

Delimiters, tags and sets

Delimiters

Tags and sets

Tags

Declearing all tags from the fst.

• LIST NAGD = Nom Acc Gen Dat ;
• LIST AGD = Acc Gen Dat ;
• LIST AD = Acc Dat ;
• LIST ND = Nom Dat ;
• LIST GENDER = Msc Fem Neu ;
• LIST NUMBER = Sg Pl ;
• @CODE

Sets

Combining tags into useful sets

Noun sets

Adjective sets

Nominal sets

Verb sets

Noun-Verb sets

Number sets

Preposition sets, taking different cases

• Union sets: ACCPREP, DATPREP, ACCDATPREP, ACCGENPREP, ACCDATGENPREP
• Intersection set: SOMEACCPREP, SOMEDATPREP, SOMEGENPREP, SOMEACCDATPREP,

Boundary sets

Case sets

These are sets of cases, not sets of prepositions choosing them.

NOTNOM, NOTDAT, etc.: Some case, but not…

Word sets

• Sem/ID when preceeded by § or currency (more should be added here)

• sub3, sub2, sub1 removes non-lexicalised compounds, the most complex ones first

Test: Go for minimal weight. This rules gives priority to lexicalised forms.

• Guess, Err/Guess removes Guess readings when other readings are available

Early and popular rules

Infinitive

• r50, IM1, IM2, InfAndInf choose Inf and at when cooccuring

• r36, r37, ImForV choosing CS over IM and Pr** at (Pr, CS, IM)

• r2 selects Pr for á when subcategorising for Dat, Acc.

A or N

Disambiguate A, N due to context.

TAD, Pron or Det

Adjective disambiguation in NP

Case disambiguation

Noun disambiguation

Conjunctions

No rules so far

Subjunctions

Verbs

Passive

-st

• bera

• hava

• muna

• síggja

vera

verða

kunna

koma

Plural

PP disambiguation

Preposition or not?

á

av

millum

móti

til

tíður

um

undir

við

Case within PP phrases

POS disambiguation

Adjectives

kalur

Pronouns

Pron Pers or Det

Det

Pron not N

Proper nouns

Specific lexemes, words

• aftan

• allur

• at

• á

• ár

• bara

• eg

• ein

• eingin

• hava

• hann

• her

• hetta

• hon

• húsi

• ið

• innan

• liggja

• men

• munandi

• niðan

• nú

• ruður

• seg

• sjalvur

• skal

• tann

• tá and tá ið

• frammanfyri and others

• um

• unglingi

• inni

• á

• ver “ver” CmpNP/None N Neu Sg Acc Def = verið

Adverbs

General adverb

Specific adverbs

akkurát

• bara

• mikið

• næstan

• her

Lexicalised adverbs.

• bara

• heldur

• líka

• væl

• saman (not samur)

Adverb verbs

Idioms

Numerals and number symbols

• NumRom in beginning of sentence

NP internal constraints

Determiner disambiguation

Specific determiners

• tað Nothing so far.

• summi

Postnominal determiner disambiguation

• Possessor number…

• Possessor case…

Definiteness disambiguation

Define definiteness based upon case concordance.

Case disambiguation

Noun disambiguation

Poss disambiguation

Ensuring case concordande within poss phrases

Number disambiguation

Coordination

Embedded clause V topicalisation

Elliptic AP as NP

P chains or not

Pronoun disambiguation

NP Coordination

VP disambiguation

V or A

V or Adv

V or N

Infinitive

Imperative

The best would be to make a corpus of imperative sentences, identify all the imperatives, and then just remove the rest.

Here come all rules selecting Imp. (so far only one)

• ImpNotSup

• CoordImp if part of Imp coordination

Then we remove the remaining ones.

• RemoveImp remove all Imp

Present participle

Supine

Present singular

Present plural

V + Refl

Past indicative

Perfect participle

Case disambiguation

Nominative

Predicative

• TadCopNom, TadCopSg for tað + copula + nominative

Subject

• NotTopAcc when sentence-initial NP cannot be Acc

• NotTopAcc2 when sentence-initial NP cannot be Acc

• NomV3 Nom, not Acc in 3rd position when Adv is topicalised

• atNom gives Nom between CC and vera.

Miscellanious

Accusative

• AccNotDatObj gives Acc not Dat for TV objects

Mær dámar

Genitive

Pronoun disambiguation

seg

Verb disambiguation

A or V

Person

Number disambiguation

Postverbal subject

Gender disamb of adjectives

Gender disamb of numerals

Case disamb of numerals

Ordinals

Coordination

• PrtNotPtc, PrtNotA gives Prt instead of participle. TODO: Find counter examples and write rules (still not found).

Adjective disambiguation outside NP

Substituting tags

Titles

CC Coordinate NPs

AFTER-section

For Apertium

• NotDat removes Dat when we did not find Dat assigners

MAPPING OF CC AND CS

Mostly we map both @CNP and @CVP, then we select @CNP, after that we remove them so @CVP remains

• CCasCNPCVP Map (@CNP @CVP) to CC

• killAllahtenotCS All occurrences of “at” are CSs.

• CS removes CS between CS

• CmpNPNone removes CmpNPNone

• Kill Sem/ID

• killAllCNP removes all remaining @CNP

• ErrOrth goes for correct forms

• NomFragment chooses Nom when no case-assigners present for Acc, Dat

• X removes readings with no syntax

---

This (part of) documentation was generated from src/cg3/disambiguator.cg3