https://github.com/ppke-nlpg/purepospy

Python wrapper for PurePos
https://github.com/ppke-nlpg/purepospy

Last synced: 3 months ago
JSON representation

Python wrapper for PurePos

• Host: GitHub
• URL: https://github.com/ppke-nlpg/purepospy
• Owner: ppke-nlpg
• License: lgpl-3.0
• Created: 2014-07-18T07:59:29.000Z (over 10 years ago)
• Default Branch: master
• Last Pushed: 2019-12-14T19:31:35.000Z (almost 5 years ago)
• Last Synced: 2024-05-18T17:58:18.879Z (6 months ago)
• Language: Java
• Size: 4.57 MB
• Stars: 1
• Watchers: 8
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

• awesome-hungarian-nlp - purepos.py

README

        
# __Warning: This repository might not contain the newest version of the source code. The development is continued at https://github.com/dlt-rilmta/purepospy__ 

---

# PurePosPy

A Python3 wrapper for [PurePos](https://github.com/ppke-nlpg/purepos).

## Requirements

- See requirements.txt.

- Model (already included) e.g. [Szeged corpus](http://rgai.inf.u-szeged.hu/index.php?lang=en&page=SzegedTreebank) with [emMorph tags](http://e-magyar.hu/en/textmodules/emmorph_codelist).

## Usage

The authors recommend using PurePOSPy in [emtsv](https://github.com/dlt-rilmta/emtsv) the new version of [e-magyar](http://www.e-magyar.hu) language processing system. This module is called *emTag*.

The PurePOS class is a Python 3 wrapper class around PurePos, which can be used for training and tagging

	```python

	>>> from purepospy import PurePOS

	>>> p = PurePOS('szeged.model')  # New, or existing file

	>>> tok = ['word', 'lemma', 'tag']

	>>> sent = [tok, tok, ...]

	>>> sentences = [[sent],[sent], ...]

	>>> p.train(sentences)  # Training, optional

	>>> p.tag_sentence('Sentence as string , tokenised .')

	Output#output_lemma#output_tag as#as_lemma#as_tag string#string_lemma#string_tag .#.#PUNCT

	```

## License

This Python wrapper, and utilities are licensed under the LGPL 3.0 license.

PurePos has its own license.