Process and Ontology Based Data Integration

Stefan Jablonski, Abdul Rehman, Bernhard Volz

Scientific research is characterized by its need for data from different heterogeneous sources. Thus methods are needed for supporting scientists in integrating data. These methods should not be based on hand-written converters. Also they should not be based on standardized ontologies and data formats since they do often not cover all characteristics of one specific scenario. We developed such a method that is based on workflow management and that allows scientists to re-use their existing (local) ontologies and formats side-by-side with standardized ontologies and formats. One characteristic of that approach is that data integration steps are hidden and not mixed up with core processing logics of a scientific analysis. Nevertheless information about syntactic and semantic transformations is contained in the data perspective of the workflow models as a matter of course; this information is interpreted at runtime by a special framework called DaltOn that then performs data transformations automatically. In this publication we explain this method in detail using examples from the ecological research.

Conference: Environmental Informatics and Industrial Ecology 2008 (EnviroInfo 2008), Lüneburg, Germany, 09.2008
Year: 09.2008

Universität Bayreuth -