[Snowball-discuss] Python port of Snowball stemmers is ready
Martin Porter
martin at porterloo.wanadoo.co.uk
Tue Jun 22 15:52:08 BST 2010
Peter,
I've corrected the snowball algorithm error (RV for R1) and made a link from
http://snowball.tartarus.org/otherlangs/index.html
to your work. I won't be using the Python versions myself -- I'm
semi-retired from computing work -- but no doubt you'll get feedback from
elsewhere!
Thanks for your interest,
Martin
At 06:53 PM 6/21/2010 +0200, Peter Stahl wrote:
>
>Hi Martin,
>
>I just want to let you know that I've finished porting twelve of your
Snowball stemmers to Python. I have made the code available on NLTK's code
repository. It can be found under
http://code.google.com/p/nltk/issues/detail?id=567.
>If you like to use it without the need for installing NLTK, then at first
you have to let the class 'SnowballStemmer' inherit from class 'object'.
Furthermore, you have to comment out the following imports:
>
>from api import *
>from nltk.corpus import stopwords
>
>Then the module should work the way I explained in the various docstrings.
However, the demo function at the end of the module will not work without
NLTK's corpora. I have not got feedback so far, so I would be happy if you
tested my module and told me about bugs etc.
>Oh, and by the way, the Python port of the German stemmer that is available
on your site does not work correctly. It contains some bugs and the code is
not of good style.
>
>Thanks and best regards,
>Peter
More information about the Snowball-discuss
mailing list