Workflow construction for service-oriented knowledge discovery

Abstract

The paper proposes a Service-oriented Knowledge Discovery (SoKD) framework and a prototype implementation named Orange4WS. To provide the proposed framework with semantics, we are using the Knowledge Discovery Ontology which defines relationships among the ingredients of knowledge discovery scenarios. It enables to reason which algorithms can be used to produce the results required by a specified knowledge discovery task, and to query the results of the knowledge discovery tasks. In addition, the ontology can also be used for automatic annotation of manually created workflows facilitating their reuse. Thus, the proposed framework provides an approach to third generation data mining: integration of distributed, heterogeneous data and knowledge resources and software into a coherent and effective knowledge discovery process. The abilities of the prototype implementation have been demonstrated on a text mining use case featuring publicly available data repositories, specialized algorithms, and third-party data analysis tools. © 2010 Springer-Verlag.