💫 Industrial-strength Natural Language Processing (NLP) in Python
Go to file
2015-02-01 18:11:44 +11:00
bin/parser * Comment training function back in 2015-01-31 16:37:57 +11:00
docs * Upd project status note 2015-02-01 16:47:33 +11:00
spacy * Upd download script 2015-02-01 13:22:23 +11:00
tests * Fix test_tag_names again 2015-02-01 16:25:03 +11:00
.gitignore * Upd gitignore 2015-01-30 16:49:44 +11:00
.travis.yml * Upd travis.yml 2015-01-31 13:50:30 +11:00
dev_setup.py * Upd dev_setup 2015-01-03 21:02:03 +11:00
fabfile.py * Fix clean command 2015-01-25 14:49:29 +11:00
LICENSE.txt * Add license file 2015-01-26 03:07:18 +11:00
MANIFEST.in * Add manifest file 2015-01-30 16:49:02 +11:00
README.md Update README.md 2015-02-01 14:47:18 +11:00
requirements.txt * Require advanced version of cymem 2015-02-01 17:04:59 +11:00
setup.py * Require advanced version of cymem 2015-02-01 17:04:59 +11:00
wordnet_license.txt * Add WordNet license file 2015-02-01 16:11:53 +11:00

spaCy

http://honnibal.github.io/spaCy

Fast, state-of-the-art natural language processing pipeline. Commercial licenses available, or use under AGPL.

  • Several bug-fixes have now been pushed to master
  • Tests fail on some platforms, including Travis CI, due to memory errors.
  • Tests pass on my local machines OSX and Ubuntu machines (for Python2.7 and Python 3.4)

The problem is likely due to non-portable usage of the Py_UNICODE data type in my Cython code, or possibly in the binary file formats of lexemes.bin, vec.bin, or the model file read by thinc.learner.LinearModel.

I'm trying to reproduce the problem. Once this is fixed and docs are updated I will push version 0.4 to PyPi.

I have a flight from Sydney to New York in 24 hours, so this problem may remain unfixed for a few days.

Supports:

  • CPython 2.7
  • CPython 3.4
  • OSX
  • Linux

Want to support:

  • Windows

Difficult to support:

  • PyPy 2.7
  • PyPy 3.4