Grid-enabled Weka

A Service-Oriented Architecture to Execute Weka’s Data Mining Algorithms on a Grid

The design goal of this prototype is to develop a system with a service-oriented architecture using Web services for data mining on the basis of the widely known data mining toolkit Weka. Web services represent state-of-the-art technology for loosely coupled systems. They are platform- and language-independent and communicate using formatted messages. The inherent complexity of the system is hidden from the end users by offering a workflow editor as the main user interface.
The adaptation of Weka to a service-oriented architecture consists of services, which provide information about the application required by the user. These services are generic in a sense that they cover all of Weka’s algorithms. This means that more than forty of Weka’s algorithms are available in the grid.

Contribution to SIMDAT
Standalone prototype; components have been integrated into the Automotive PM24 prototype

After successful integration of the Weka data mining tool the statistic package R will also become Grid-enabled. It can be used as services in the context of SIMDAT, on the Auto demonstrator for example.

Modified Weka components and Gria interface available from Fraunhofer IAIS

Dr. Jörg Kindermann
Fraunhofer Institute IAIS