283 lines
10 KiB
Plaintext
283 lines
10 KiB
Plaintext
# Mapping from JOS PoS to UD 2.0 PoS
|
||
# Kaja Dobrovoljc, Tomaž Erjavec, Simon Krek
|
||
# 2019-02-04
|
||
#
|
||
#Prio Lemma Category Feats Deps ->PoS-UD #Comment
|
||
#-------------------------------------------------------------------------------------------------------
|
||
3 * Noun Type=common * NOUN
|
||
3 * Noun Type=proper * PROPN
|
||
|
||
3 * Verb * * VERB
|
||
|
||
2 * Verb Type=auxiliary * AUX #This is one can in fact also be VERB, but this has to be determined by some other means
|
||
|
||
3 * Adjective * * ADJ
|
||
|
||
3 * Adverb * * ADV
|
||
1 četrt Adverb * * DET
|
||
1 čimmanj Adverb * * DET
|
||
1 čimveč Adverb * * DET
|
||
1 dosti Adverb * * DET
|
||
1 dovolj Adverb * * DET
|
||
1 enako Adverb * * ADV
|
||
1 enormno Adverb * * DET
|
||
1 ful Adverb * * ADV
|
||
1 koliko Adverb * * DET
|
||
1 majčkeno Adverb * * DET
|
||
1 maksimalno Adverb * * ADV
|
||
1 malce Adverb * * ADV
|
||
1 malo Adverb * * DET
|
||
1 manj Adverb * * DET
|
||
1 minimalno Adverb * * ADV
|
||
1 mnogo Adverb * * DET
|
||
1 najmanj Adverb * * ADV
|
||
1 največ Adverb * * DET
|
||
1 nekaj Adverb * * DET
|
||
1 nekoliko Adverb * * ADV
|
||
1 nemalo Adverb * * ADV
|
||
1 nešteto Adverb * * DET
|
||
1 nič Adverb * * ADV
|
||
1 ničkoliko Adverb * * DET
|
||
1 obilo Adverb * * DET
|
||
1 ogromno Adverb * * DET
|
||
1 par Adverb * * DET
|
||
1 pol Adverb * * DET
|
||
1 polno Adverb * * ADV
|
||
1 precej Adverb * * ADV
|
||
1 premalo Adverb * * ADV
|
||
1 premnogo Adverb * * DET
|
||
1 preveč Adverb * * DET
|
||
1 toliko Adverb * * DET
|
||
1 veliko Adverb * * DET
|
||
1 več Adverb * * DET
|
||
1 večidel Adverb * * ADV
|
||
1 vse Adverb * * ADV
|
||
1 zadosti Adverb * * ADV
|
||
|
||
##All Pronouns should be explicitly defined
|
||
##But are not because of jos1M wrong lemmatisations for e.g. "ti", "te" etc.
|
||
3 * Pronoun * * PRON
|
||
##2 * Pronoun Type=demonstrative * DET
|
||
##2 * Pronoun Type=possessive * DET
|
||
1 bogsigavedikakšen Pronoun Type=indefinite * DET
|
||
1 bogvedikaj Pronoun Type=indefinite * PRON
|
||
1 bogvedikateri Pronoun Type=indefinite * DET
|
||
1 bogvekaj Pronoun Type=indefinite * PRON
|
||
1 bogvekakšen Pronoun Type=indefinite * DET
|
||
1 bogvekateri Pronoun Type=indefinite * DET
|
||
1 bogvekolik Pronoun Type=indefinite * DET
|
||
1 bogvekolikšen Pronoun Type=indefinite * DET
|
||
1 čezme Pronoun Type=personal * PRON
|
||
1 čezse Pronoun Type=reflexive * PRON
|
||
1 čigar Pronoun Type=relative * DET
|
||
1 čigarkoli Pronoun Type=relative * DET
|
||
1 čigarsižebodi Pronoun Type=relative * DET
|
||
1 čigav Pronoun Type=interrogative * DET
|
||
1 čigaver Pronoun Type=relative * DET
|
||
1 čigaverkoli Pronoun Type=relative * DET
|
||
1 čigavršen Pronoun Type=relative * DET
|
||
1 čigavršnji Pronoun Type=relative * DET
|
||
1 enak Pronoun Type=indefinite * DET
|
||
1 enaki Pronoun Type=indefinite * DET
|
||
1 enakšen Pronoun Type=indefinite * DET
|
||
1 isti Pronoun Type=indefinite * DET
|
||
1 jaz Pronoun Type=personal * PRON
|
||
1 jest Pronoun Type=personal * PRON
|
||
1 kaj Pronoun Type=interrogative * PRON
|
||
1 kak Pronoun Type=interrogative * DET
|
||
1 kakov Pronoun Type=interrogative * DET
|
||
1 kakošen Pronoun Type=interrogative * DET
|
||
1 kakršen Pronoun Type=relative * DET
|
||
1 kakršenkoli Pronoun Type=relative * DET
|
||
1 kakršensižebodi Pronoun Type=relative * DET
|
||
1 kakšen Pronoun Type=interrogative * DET
|
||
1 kar Pronoun Type=relative * PRON
|
||
1 karkoli Pronoun Type=relative * PRON
|
||
1 karsibodi Pronoun Type=relative * PRON
|
||
1 karsižebodi Pronoun Type=relative * PRON
|
||
1 kateri Pronoun Type=interrogative * DET
|
||
1 katerikoli Pronoun Type=relative * DET
|
||
1 katerisibodi Pronoun Type=relative * DET
|
||
1 kdo Pronoun Type=interrogative * PRON
|
||
1 kdor Pronoun Type=relative * PRON
|
||
1 kdorkoli Pronoun Type=relative * PRON
|
||
1 kdorsibodi Pronoun Type=relative * PRON
|
||
1 kdorsižebodi Pronoun Type=relative * PRON
|
||
1 kdovekaj Pronoun Type=indefinite * PRON
|
||
1 kdovekak Pronoun Type=indefinite * DET
|
||
1 kdovekakšen Pronoun Type=indefinite * DET
|
||
1 kdovekateri Pronoun Type=indefinite * DET
|
||
1 kdovekdo Pronoun Type=indefinite * PRON
|
||
1 kdovekolik Pronoun Type=indefinite * DET
|
||
1 koji Pronoun Type=interrogative * DET
|
||
1 kolik Pronoun Type=interrogative * DET
|
||
1 kolik Pronoun Type=indefinite * DET
|
||
1 koliker Pronoun Type=interrogative * DET
|
||
1 kolikršen Pronoun Type=relative * DET
|
||
1 kolikšen Pronoun Type=interrogative * DET
|
||
1 malokaj Pronoun Type=indefinite * PRON
|
||
1 malokak Pronoun Type=indefinite * DET
|
||
1 malokakšen Pronoun Type=indefinite * DET
|
||
1 malokateri Pronoun Type=indefinite * DET
|
||
1 malokdo Pronoun Type=indefinite * PRON
|
||
1 marsikaj Pronoun Type=indefinite * PRON
|
||
1 marsikak Pronoun Type=indefinite * DET
|
||
1 marsikakšen Pronoun Type=indefinite * DET
|
||
1 marsikateri Pronoun Type=indefinite * DET
|
||
1 marsikdo Pronoun Type=indefinite * PRON
|
||
1 marsičigav Pronoun Type=indefinite * DET
|
||
1 medme Pronoun Type=personal * PRON
|
||
1 medse Pronoun Type=reflexive * PRON
|
||
1 mnog Pronoun Type=indefinite * DET
|
||
1 mnogokaj Pronoun Type=indefinite * PRON
|
||
1 mnogokateri Pronoun Type=indefinite * DET
|
||
1 mnogokdo Pronoun Type=indefinite * PRON
|
||
1 moj Pronoun Type=possessive * DET
|
||
1 nadme Pronoun Type=personal * PRON
|
||
1 nadse Pronoun Type=reflexive * PRON
|
||
1 najin Pronoun Type=possessive * DET
|
||
1 name Pronoun Type=personal * PRON
|
||
1 nase Pronoun Type=reflexive * PRON
|
||
1 naš Pronoun Type=possessive * DET
|
||
1 negdo Pronoun Type=indefinite * PRON
|
||
1 nek Pronoun Type=indefinite * DET
|
||
1 nekaj Pronoun Type=indefinite * PRON
|
||
1 nekak Pronoun Type=indefinite * DET
|
||
1 nekakov Pronoun Type=indefinite * DET
|
||
1 nekakšen Pronoun Type=indefinite * DET
|
||
1 nekateri Pronoun Type=indefinite * DET
|
||
1 nekdo Pronoun Type=indefinite * PRON
|
||
1 neki Pronoun Type=indefinite * DET
|
||
1 nekolik Pronoun Type=indefinite * DET
|
||
1 nekolikšen Pronoun Type=indefinite * DET
|
||
1 nekolikšnji Pronoun Type=indefinite * DET
|
||
1 nekov Pronoun Type=indefinite * DET
|
||
1 nekšen Pronoun Type=indefinite * DET
|
||
1 nevemkakšen Pronoun Type=indefinite * DET
|
||
1 nihče Pronoun Type=negative * PRON
|
||
1 nikak Pronoun Type=negative * DET
|
||
1 nikakršen Pronoun Type=negative * DET
|
||
1 nikakšen Pronoun Type=negative * DET
|
||
1 nikdo Pronoun Type=negative * PRON
|
||
1 nikogaršen Pronoun Type=negative * DET
|
||
1 nikogaršnji Pronoun Type=negative * DET
|
||
1 nič Pronoun Type=negative * PRON
|
||
1 njegov Pronoun Type=possessive * DET
|
||
1 njen Pronoun Type=possessive * DET
|
||
1 njihen Pronoun Type=possessive * DET
|
||
1 njihnji Pronoun Type=possessive * DET
|
||
1 njihov Pronoun Type=possessive * DET
|
||
1 njun Pronoun Type=possessive * DET
|
||
1 nobeden Pronoun Type=negative * PRON
|
||
1 noben Pronoun Type=negative * DET
|
||
1 oba Pronoun Type=general * DET
|
||
1 obadva Pronoun Type=general * PRON
|
||
1 obme Pronoun Type=personal * PRON
|
||
1 oboj Pronoun Type=general * DET
|
||
1 obojen Pronoun Type=general * DET
|
||
1 obse Pronoun Type=reflexive * PRON
|
||
1 on Pronoun Type=personal * PRON
|
||
1 oni Pronoun Type=demonstrative * DET
|
||
1 onile Pronoun Type=demonstrative * PRON
|
||
1 podme Pronoun Type=personal * PRON
|
||
1 podse Pronoun Type=reflexive * PRON
|
||
1 pome Pronoun Type=personal * PRON
|
||
1 predme Pronoun Type=personal * PRON
|
||
1 predse Pronoun Type=reflexive * PRON
|
||
1 premarsikateri Pronoun Type=indefinite * DET
|
||
1 premnog Pronoun Type=indefinite * DET
|
||
1 prenekaj Pronoun Type=indefinite * PRON
|
||
1 prenekateri Pronoun Type=indefinite * DET
|
||
1 prenekdo Pronoun Type=indefinite * PRON
|
||
1 redkokateri Pronoun Type=indefinite * DET
|
||
1 redkokdo Pronoun Type=indefinite * PRON
|
||
1 se Pronoun Type=reflexive * PRON
|
||
1 skozme Pronoun Type=personal * PRON
|
||
1 skozse Pronoun Type=reflexive * PRON
|
||
1 svoj Pronoun Type=reflexive * DET
|
||
1 ta Pronoun Type=demonstrative * DET
|
||
1 tadva Pronoun Type=demonstrative * PRON
|
||
1 taisti Pronoun Type=demonstrative * DET
|
||
1 tak Pronoun Type=demonstrative * DET
|
||
1 takisti Pronoun Type=demonstrative * DET
|
||
1 takle Pronoun Type=demonstrative * DET
|
||
1 takov Pronoun Type=demonstrative * DET
|
||
1 takošen Pronoun Type=demonstrative * DET
|
||
1 takšen Pronoun Type=demonstrative * DET
|
||
1 takšenle Pronoun Type=demonstrative * DET
|
||
1 tale Pronoun Type=demonstrative * DET
|
||
1 talele Pronoun Type=demonstrative * DET
|
||
1 teu Pronoun Type=personal * PRON
|
||
1 ti Pronoun Type=personal * PRON
|
||
1 tisti Pronoun Type=demonstrative * DET
|
||
1 tistile Pronoun Type=demonstrative * DET
|
||
1 tolik Pronoun Type=demonstrative * DET
|
||
1 toliker Pronoun Type=demonstrative * DET
|
||
1 tolikšen Pronoun Type=demonstrative * DET
|
||
1 tolikšnji Pronoun Type=demonstrative * DET
|
||
1 toti Pronoun Type=demonstrative * DET
|
||
1 tvoj Pronoun Type=possessive * DET
|
||
1 un Pronoun Type=demonstrative * DET
|
||
1 vajin Pronoun Type=possessive * DET
|
||
1 vame Pronoun Type=personal * PRON
|
||
1 vase Pronoun Type=reflexive * PRON
|
||
1 vaš Pronoun Type=possessive * DET
|
||
1 ves Pronoun Type=general * DET
|
||
1 vsak Pronoun Type=general * DET
|
||
1 vsakateri Pronoun Type=general * DET
|
||
1 vsakdo Pronoun Type=general * PRON
|
||
1 vsakogaršen Pronoun Type=general * DET
|
||
1 vsakogaršnji Pronoun Type=general * DET
|
||
1 vsakršen Pronoun Type=general * DET
|
||
1 vsakteri Pronoun Type=general * DET
|
||
1 zame Pronoun Type=personal * PRON
|
||
1 zase Pronoun Type=reflexive * PRON
|
||
|
||
3 * Numeral Form=digit * NUM
|
||
3 * Numeral Form=roman * NUM
|
||
3 * Numeral Form=letter|Type=special * NUM
|
||
3 * Numeral Form=letter|Type=cardinal * NUM
|
||
2 * Numeral Form=letter|Type=ordinal * ADJ
|
||
1 drug Numeral Form=letter|Type=pronominal * ADJ
|
||
1 en Numeral Form=letter|Type=pronominal * NUM
|
||
1 *en Numeral Form=letter|Type=special * ADJ #enojen, dvojen
|
||
1 eden Numeral Form=letter|Type=pronominal * NUM #Dodal E.T.
|
||
|
||
3 * Adposition * * ADP #MULTEXT-East name
|
||
3 * Preposition * * ADP #JOS name
|
||
|
||
3 * Conjunction Type=coordinating * CCONJ
|
||
3 * Conjunction Type=subordinating * SCONJ
|
||
|
||
3 * Particle * * PART
|
||
|
||
3 * Interjection * * INTJ
|
||
|
||
3 * Abbreviation * * X
|
||
|
||
3 * Residual * * X
|
||
2 * Residual Type=web * SYM
|
||
2 * Residual Type=emo * SYM
|
||
2 * Residual Type=hashtag * SYM #Better mapping?
|
||
2 * Residual Type=at * SYM #Better mapping?
|
||
2 * Residual Type=foreign * X #Better mapping?
|
||
|
||
3 * Punctuation * * PUNCT
|
||
1 # Punctuation * * SYM
|
||
1 % Punctuation * * SYM
|
||
1 & Punctuation * * SYM
|
||
1 < Punctuation * * SYM
|
||
1 > Punctuation * * SYM
|
||
1 + Punctuation * * SYM
|
||
1 = Punctuation * * SYM
|
||
1 ° Punctuation * * SYM
|
||
1 × Punctuation * * SYM
|
||
1 ÷ Punctuation * * SYM
|
||
1 $ Punctuation * * SYM
|
||
1 @ Punctuation * * SYM
|
||
1 µ Punctuation * * SYM
|
||
1 © Punctuation * * SYM
|
||
1 § Punctuation * * SYM
|
||
1 € Punctuation * * SYM
|
||
1 £ Punctuation * * SYM
|