SETUP

Setup:

(1) You need Python 2.5 or later. Python 3 does not seem to work.
(2) To use SVM models, you'll need SVM Light (http://svmlight.joachims.org/)
      - once installed, you'll need to modify sbd.py slightly:
        at the top, change the paths to SVM_LEARN and SVM_CLASSIFY to point
        to the files you've installed.

-----

Example calls (if python defaults to version 2.4, you'll need to specify
               python2.5):

(show command line options)
python sbd.py -h

(split sentences in sample.txt using the provided Naive Bayes model)
python sbd.py -m model_nb sample.txt

(now using the provided SVM model)
python sbd.py -m model_svm sample.txt

(now keeping tokenized output)
python sbd.py -m model_nb -t sample.txt

(now writing output to sample.sent)
python sbd.py -m model_nb -t sample.txt -o sample.sent

-----

Note about SVM_LIGHT:

The provided SVM model was built with SVM_LIGHT version 6.01. It seems that
SVM_CLASSIFY requires a matching version or it will crash. So, you can either
try to use version 6.01, or you can make the following quick fix:

open model_svm/svm_model
change the first line from:
SVM-light Version V6.02
to whatever your version is.