MultiCOMBO
Multilingual POS-Tagger and Dependency-Parser with COMBO-pytorch and spaCy
Basic usage
>>> import multicombo
>>> nlp=multicombo.load()
>>> doc=nlp('Who plays "La vie en rose"?')
>>> print(multicombo.to_conllu(doc))
# text = Who plays "La vie en rose"?
1 Who _ PRON _ PronType=Int 2 nsubj _ Translit=who
2 plays _ VERB _ Mood=Ind|Number=Sing|Person=3|Tense=Pres|VerbForm=Fin 0 root _ _
3 " _ PUNCT _ _ 5 punct _ SpaceAfter=No
4 La _ DET _ Definite=Def|Gender=Fem|Number=Sing|PronType=Art 5 det _ Translit=la
5 vie _ NOUN _ Gender=Fem|Number=Sing 2 obj _ _
6 en _ ADP _ _ 7 case _ _
7 rose _ NOUN _ Number=Sing 5 nmod _ SpaceAfter=No
8 " _ PUNCT _ _ 5 punct _ SpaceAfter=No
9 ? _ PUNCT _ _ 2 punct _ SpaceAfter=No
>>> import deplacy
>>> deplacy.render(doc)
Who PRON <════════════╗ nsubj
plays VERB ═══════════╗═╝═╗ ROOT
" PUNCT <══════╗ ║ ║ punct
La DET <════╗ ║ ║ ║ det
vie NOUN ═══╗═╝═╝═╗<╝ ║ obj
en ADP <╗ ║ ║ ║ case
rose NOUN ═╝<╝ ║ ║ nmod
" PUNCT <════════╝ ║ punct
? PUNCT <══════════════╝ punct
>>> deplacy.serve(doc)
http://127.0.0.1:5000
multicombo.load(lang="xx") loads spaCy Language pipeline with bert-base-multilingual-cased and spacy.lang.xx.MultiLanguage tokenizer. Other language specific tokenizers can be loaded with the option lang, while several languages require additional packages:
-
lang="ja"Japanese requires SudachiPy and SudachiDict-core. -
lang="th"Thai requires PyThaiNLP. -
lang="vi"Vietnamese requires pyvi.
Installation for Linux
pip3 install multicombo --userInstallation for Cygwin64
Make sure to get python37-devel python37-pip python37-cython python37-numpy python37-cffi gcc-g++ mingw64-x86_64-gcc-g++ gcc-fortran git curl make cmake libopenblas liblapack-devel libhdf5-devel libfreetype-devel libuv-devel packages, and then:
curl -L https://raw.githubusercontent.com/KoichiYasuoka/UniDic-COMBO/master/cygwin64.sh | sh
pip3.7 install multicomboInstallation for Jupyter Notebook (Google Colaboratory)
!pip install multicomboTry notebook for Google Colaboratory.
