Evan Klitzke wrote:
> What frameworks are there available for doing pattern classification?
> I'm generally interested in the problem of mapping some sort of input
> to one or more categories. For example, I want to be able to solve
> problems like taking text and applying one or more tags to it like
> "romance", "horror", "poetry", etc. This isn't really my research
> specialty, but my understanding is that Bayesian classifiers are
> generally used for problems like this.
In fact, a wide variety of classifiers are used in text classification,
including Bayesian approaches, support vector machines, conditional
random fields, etc.
> Are there any other frameworks I should be aware of?
I have used (but not recently) Orange:
http://www.ailab.si/orange
I haven't used, but have been meaning to try, PyML:
http://pyml.sourceforge.net/
A more recent addition (whose documentation needs work) is:
http://montepython.sourceforge.net/
And here's a Summer of Code project to build an ML library:
http://projects.scipy.org/scipy/scipy/wiki/MachineLearning
These are all general-purpose machine learning frameworks. So they can
be applied to pretty much any classification problem (including the text
classification problems you're looking at). You just need to pick out a
set of relevant features to describe your data, and feed those features
along with your chosen labels to a machine learning algorithm.
STeVe
--
http://mail.python.org/mailman/listinfo/python-list
