KeplerWeka represents a machine learning module for the scientific workflow
engine Kepler (http://www.kepler-project.org/). It uses the WEKA machine
learning workbench (http://www.cs.waikato.ac.nz/ml/weka/) for all the machine
learning tasks it offers.
KeplerWeka is released under the GNU General Public License 2.
See module-info/COPYING for details on the license.
In order to get KeplerWeka running, you need to modify your Kepler installation
or source tree. It is recommended to backup your current Kepler before
proceeding with the installation of KeplerWeka.
1.1 Installation using KeplerWeka release
Previously installed versions of KeplerWeka should be removed first, of course.
Deleting the "keplerweka" sub-directory in the Kepler installation directory
should be sufficient for that.
In order to install KeplerWeka, simply decompress the ZIP archive into your
Kepler installation directory. NB this will overwrite any currently active
1.2 Installation from source code
The following sections cover the various steps that you need to perform in
order to get KeplerWeka installed using source code obtain from a source
archive or from subversion directly.
1.2.1 KeplerWeka Suite
If you are using a source archive, simply decompress it into your Kepler
source tree directory. Otherwise, if you are obtaining the source code
directly from subversion, run the following command in your Kepler source
svn co https://keplerweka.svn.sourceforge.net/svnroot/keplerweka/trunk keplerweka
The suite is built using the provided ant build file (tested only under Linux).
Issue the following command from within the "keplerweka" module directory:
KeplerWeka extends the Kepler ontologies a bit, making it easier to find
the actors. Minor changes to a Kepler configuration file are necessary.
Copy the following files:
Into the following directory:
Edit the following file:
And add the content of the following file to it (inside the
In order to display KeplerWeka's custom icons, you need to updated some
Kepler configuration files, unfortunately. KeplerWeka is fully functional
without the icons, but the actors will be easier to distinguish with their
proper icons installed.
Add the content of:
to the following Kepler configuration file (inside the <config>...</config>
The weka.jar library in the lib/jar directory can be replaced with any WEKA jar
of the developer version that is later than revision 6896 (>= 3.7.3).
Packages that were installed through WEKA's package manager, are accessible
in KeplerWeka as well. The partialLeastSquares package is already available
in the lib/jar directory, as the setup generator uses the PLS filter in its
If you obtained the source code of KeplerWeka, then you can include it in
your Kepler Eclipse environment. First, copy the file ".classpath.default"
to ".classpath". Then, you can import the project in Eclipse.
$Revision: 197 $