This code creates a map from wordform to part-of-speech using UD data.

To use the ud2pos tagger, first install this repository and then instantiate the UdTagger class.

To install the code in this repository run:

$ pip install -e .

To use the ud2pos tagger, add the following to your script:

import ud2pos
language = 'english'
tagger = ud2pos.UdTagger(language)
print('Pos tag is: %s' % tagger('Hi'))

This pos tagger will return 'UNK' for words not in the universal dependencies.

To process ud data for new languages, run the following commands:

Create a conda environment with

$ conda env create -f environment.yml

You can easily download UD data with the following command

$ make get_ud

You can then get the embeddings for a language with command

$ make process LANGUAGE=<language>

As languages, you should be able to experiment on any in UD. For instance: 'english'; 'czech'; 'basque'; 'finnish'; 'turkish'; 'arabic'; 'japanese'; 'tamil'; 'korean'; 'marathi'; 'urdu'; 'telugu'; 'indonesian'.

Upgrade the repository version in setup.py. Then run:

$ python setup.py sdist
$ twine upload dist/*

To ask questions or report problems, please open an issue.