==================
practNLPTools-lite
==================
Creating practNLPTools in lite mode.[ get the old coding in `devbranch`_ or old stable code `properbranch`_]
|Author|
|Build Status| - on click this built this might take you to build of
`practNLPTools`_ which is testing ground for this repository so don’t
worry.
Practical Natural Language Processing Tools for Humans.
practNLPTools is a pythonic library over `SENNA`_ and Stanford
Dependency Extractor.
.. |pypi status| image:: https://img.shields.io/pypi/v/practNLPTools-lite.svg
:target: https://pypi.python.org/pypi/pntl
.. |travis status| image:: https://img.shields.io/travis/jawahar273/practNLPTools-lite.svg
:target: https://travis-ci.org/jawahar273/practNLPTools-lite
.. |doc status| image:: https://readthedocs.org/projects/pntl/badge/?version=latest
:target: https://pntl.readthedocs.io/en/latest/?badge=latest
:alt: Documentation Status
.. |dep status| image:: https://pyup.io/repos/github/jawahar273/practNLPTools-lite/shield.svg
:target: https://pyup.io/repos/github/jawahar273/practNLPTools-lite/
:alt: Updates
.. |blocker status| image:: https://pyup.io/repos/github/jawahar273/practNLPTools-lite/python-3-shield.svg
:target: https://pyup.io/repos/github/jawahar273/practNLPTools-lite/
:alt: Python 3
.. |Wercker status| image:: https://app.wercker.com/status/758bf4fa0e3bb9066d118385ee4aac1f/s/master
:target: https://app.wercker.com/project/byKey/758bf4fa0e3bb9066d118385ee4aac1f
+-----------------+-----------------+
| name | status |
+=================+=================+
| Wercker status | |Wercker status||
+-----------------+-----------------+
| PyPi | |pypi status||
+-----------------+-----------------+
| travis | |travis status||
+-----------------+-----------------+
| Documentation | |doc status||
+-----------------+-----------------+
| dependency | |dep status||
+-----------------+-----------------+
| blocker Pyupbot | |blocker status||
+-----------------+-----------------+
| FOSSA | |FOSSA Status||
+-----------------+-----------------+
* Documentation: https://pntl.readthedocs.io
QuickStart
==========
Downlarding Stanford Parser JAR
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
To downlard the stanford-parser from github automatically and placing them inside the install direction.
.. code:: bash
pntl -I true
# downlards required file from github.
Running Predefine Examples Sentences
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
To run exiting example for batch(which has more than one list of examples).
.. code:: bash
pntl -SE home/user/senna -B true
To run predefine example for one sentence.
.. code:: bash
pntl -SE home/user/senna
Running user given sentence
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
To run user given example using `-S` is
.. code:: bash
pntl -SE home/user/senna -S 'I am gonna make him an offer he can not refuse.'
Functionality
=============
- Semantic Role Labeling.
- Syntactic Parsing.
- Part of Speech Tagging (POS Tagging).
- Named Entity Recognisation (NER).
- Dependency Parsing.
- Shallow Chunking.
- Skip-gram(in-case).
- find the senna path if is install in the system.
- stanford parser and depPaser file into installed direction.
Future work
===========
- tag2file(new)
- creating depParser for corresponding os environment
- custome input format for stanford parser insted of tree format
Features
========
#. Fast: `SENNA`_ is written is C. So it is Fast.
#. We use only dependency Extractor Component of Stanford Parser, which
takes in Syntactic Parse from SENNA and applies dependency
Extraction. So there is no need to load parsing models for Stanford
Parser, which takes time.
#. Easy to use.
#. Platform Supported - Windows, Linux and Mac
#. Automatic finds stanford parsing jar if it is present in install path[pntl].
.. note::
SENNA pipeline has a fixed maximum size of the sentences that it
can read. By default it is 1024 token/sentence. If you have larger
sentences, changing the MAX\_SENTENCE\_SIZE value in SENNA\_main.c should beconsidered and your system specific binary should be rebuilt. Otherwise this could introduce misalignment errors.
Installation
============
Requires:
~~~~~~~~~
A computer with 500mb memory, Java Runtime Environment (1.7
preferably, works with 1.6 too, but didnt test.) installed and python.
Linux:
```````
run:
::
sudo python setup.py install
windows:
````````
run this commands as administrator:
::
python setup.py install
Bench Mark comparsion
=====================
By using the ``time`` command in ubuntu on running the ``testsrl.py`` on
this `link`_ and along with ``tools.py`` on ``pntl``
.. _link: https://github.com/jawahar273/SRLTagger
+-----------------+-----------------+-----------------+
| | pntl | NLTK-senna |
+=================+=================+=================+
| at fist run | | |
+-----------------+-----------------+-----------------+
| | real 0m1.674s | real 0m2.484s |
+-----------------+-----------------+-----------------+
| | user 0m1.564s | user 0m1.868s |
+-----------------+-----------------+-----------------+
| | sys 0m0.228s | sys 0m0.524s |
+-----------------+-----------------+-----------------+
| at second run | | |
+-----------------+-----------------+-----------------+
| | real 0m1.245s | real 0m3.359s |
+-----------------+-----------------+-----------------+
| | user 0m1.560s | user 0m2.016s |
+-----------------+-----------------+-----------------+
| | sys 0m0.152s | sys 0m1.168s |
+-----------------+-----------------+-----------------+
.. note::
this bench mark may differt accouding to system’s working and to restult present here is exact same result in my system ububtu 4Gb RAM
and i3 process. If I find another good benchmark techinque then I will
change to it.
.. _practNLPTools: https://github.com/jawahar273/practNLPTools-lite
.. _SENNA: http://ronan.collobert.com/senna/
.. _properbranch: https://github.com/jawahar273/practNLPTools-lite/tree/pyup-update-pytest-3.2.2-to-3.2.3
.. _devbranch: https://github.com/jawahar273/practNLPTools-lite/tree/dev
.. |Author| image:: https://img.shields.io/badge/Author-jawahar-blue.svg
.. |Python-version-3| image:: https://img.shields.io/badge/Py-version-Python--3.5-green.svg
.. |Build Status| image:: https://travis-ci.org/jawahar273/practNLPTools.svg?branch=master
:target: https://travis-ci.org/jawahar273/practNLPTools
.. |FOSSA Status| image:: https://app.fossa.io/api/projects/git%2Bhttps%3A%2F%2Fgithub.com%2Fjawahar273%2FpractNLPTools-lite.svg?type=small
:target: https://app.fossa.io/projects/git%2Bhttps%3A%2F%2Fgithub.com%2Fjawahar273%2FpractNLPTools-lite?ref=badge_small
.. Features
.. --------
.. * TODO
Credits
=======
This package was created with Cookiecutter_ and the `audreyr/cookiecutter-pypackage`_ project template.
.. _Cookiecutter: https://github.com/audreyr/cookiecutter
.. _`audreyr/cookiecutter-pypackage`: https://github.com/audreyr/cookiecutter-pypackage
=======
History
=======
0.2.3-beta
----------
* Proper release for PyPI version.
0.2.0 4-alpha
------------------
* Marking standard tools for `pntl`.
0.1.1 (2017-09-17)
------------------
* Planing to release on PyPI.
Metadata
Release files for pntl 0.2.4
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | |
|---|---|---|---|---|
| pntl-0.2.4.tar.gz | 2.5 MB | Without provenance | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| pntl-0.2.4-py2.py3-none-any.whl | Python 3, Python 2 | none | any | Details |
Total release size: 4.9 MB
Release files / pntl-0.2.4.tar.gz
| Download URL | pntl-0.2.4.tar.gz |
|---|---|
| Size | 2.5 MB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
b258746d667714c292f09a66d3b8919c6b4ab797014edcced73c9937d7ac165a
|
|
BLAKE2b-256 checksum How to use checksums |
dba364ab82e23b5cca360d83134ed43ed48bbf826b42fe91d35dfb6b0de67a3c
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
Release files / pntl-0.2.4-py2.py3-none-any.whl
| Download URL | pntl-0.2.4-py2.py3-none-any.whl |
|---|---|
| Size | 2.5 MB |
| Tags | Python 2 Python 3 |
|
SHA-256 checksum How to use checksums |
44939d0404fbef5bd016d72355956c6ac457d2ea33fab6a951ba9ee0500128cd
|
|
BLAKE2b-256 checksum How to use checksums |
a4952d76e2c2e6f08949e7d8ba6ee56e4f2233f51c129b487bd4008f0cf7f833
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
Security for pntl 0.2.4
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.