CRF
CRF copied to clipboard
witgo
CRF is a Java implementation of Conditional Random Fields, an algorithm for learning from labeled sequences of examples. It also includes an implementation of Maximum Entropy learning.
You will need jdk1.4 or above and ant to compile.
Make sure to set the classpath by running "settings.sh" after setting CRF_HOME to the directory where you have unpackaged this distribution.
A sample dataset for the segmentation problem is available under samples Run it as
"java iitb.Segment.Segment train -f samples/us50.conf" "java iitb.Segment.Segment test -f samples/us50.conf" "java iitb.Segment.Segment calc -f samples/us50.conf" (report performance statistics)
or all together with
"java iitb.Segment.Segment all -f samples/us50.conf"
Files
build/ -- Dir containing all class files
build.xml -- XML file to build the code using ANT
doc/ -- Documentation for the code (It includes Java API, introduction and FAQs - all in the HTML form).
lib/ -- All required libraries are kept here.
LICENSE.txt -- Licencse agreeement.
README -- This file.
samples/ -- Sample configuration and data files.
settings.sh -- Script to set up the package.
src/ -- Source code.
third-party-license.txt -- Third party license.
witgo