I've cut a release of scalanlp based on Scalala 0.4 and Scala 2.8.
There are a lot of smaller changes, too. I'm calling it 0.3.1, and you
can get them here:
http://repo.scalanlp.org/repo/org/scalanlp/scalanlp-data_2.8.0/0.3.1/
http://repo.scalanlp.org/repo/org/scalanlp/scalanlp-learn_2.8.0/0.3.1/
ScalaNLP-Core consists of two pieces. ScalaNLP Data contains of
support classes for data and text processing and computation
pipelining. ScalaNLP Learn includes commonly used learning and
optimization algorithms, such as L-BFGS, a logistic classifier, and an
SVM. It also contains statistical distributions and sampling routines.
The goals for the next release include:
Data:
* Better handling of configuration files and command line arguments
* Better sentence and word segmentation
Learn:
* A better CRF interface, along with some good demo code (an NER
system or a POS tagger maybe?)
* Kernelized algorithms for a few things.
* More optimization routines, and hopefully a trait-heavy
implementation that will allow mixing and matching of different
components.
FST:
* I hope to clean up and document the finite-state library I've been working on.