Python Bayes implementations
by Kevin Dangoor
A couple of bookmarks of Python naive Bayes implementations for myself: Divmod Reverend is licensed under LGPL and the ever popular SpamBayes is licensed under the Python license.
A couple of bookmarks of Python naive Bayes implementations for myself: Divmod Reverend is licensed under LGPL and the ever popular SpamBayes is licensed under the Python license.
Here’s an RSS parser that is liberally licensed:
http://www.mnot.net/python/RSS.py
I combined Python Bayes with a lexcical database (Wordnet) and got some cool results.
You can do stuff like. Train(‘bad’,'crime’) and then Guess(‘murder’) returns ‘bad’.
Its pretty neat and I have released it to the world via a web interface. With enough training from the blogosphere it could computer our collective conscience.
Check it out: jrhicks.net/reverend
Nifty idea. I have a feeling that it’ll be difficult to get everyone to agree on what’s good or bad, though
Also interesting looking is the NLTK toolkit that includes different kinds of classifiers:
http://nltk.sourceforge.net/
Mostly meant for education, but could have practical value.
Too bad the link to the Reverend is 404-ing
Reverend can be found here:
http://sourceforge.net/project/showfiles.php?group_id=81259