pitekus at gmail.com
Mon May 23 10:16:00 BST 2011
You should take a look at wapiti: http://wapiti.limsi.fr/
It supports both logistic regression (aka Maxent) and Conditional
Random Fields, and implements several estimation algorithms. Written
in C and has a BSD style license.
I used it a bit and it is easy to use and fast.
You can see how it performs on the MLComp site: http://mlcomp.org/programs/771
This site is in general good for playing around with ML algorithms and datasets.
My own perceptron sequence tagger written in Haskell also performs
decently on many MLComp datasets:
Hope this helps,
On Mon, May 23, 2011 at 00:26, Rogan Creswick <creswick at gmail.com> wrote:
> Is anyone aware of any BSD/Apache (or "free-er") licensed maxent
> libraries/packages for haskell or c (or other ffi-compatible
> language..)? I'd like to recreate some of the OpenNLP classifiers,
> which rely heavily on maxent (I haven't found suitable tokenizers /
> sentence splitters / POS taggers / NE-recognisers for Haskell yet..).
> There is a c++ implementation of maxent here:
> http://www-tsujii.is.s.u-tokyo.ac.jp/~tsuruoka/maxent/ but the license
> doesn't make allowances for commercial distribution/use, which is a
> non-starter. (The GPL is also probably out, for similar reasons.)
> NLP mailing list
> NLP at projects.haskell.org
More information about the NLP