Abstract

Information integration within enterprises is hindered by differences in software and hardware platforms and by syntactic and semantic differences in the schemas of the various heterogeneous data sources. This is a well-known problem in the area of Enterprise Application Integration (EAI), where many applications have been developed for the purpose of information integration. Most current tools, however, only address the problems of (soft- and hardware) platform and syntactic heterogeneity. The problem of semantic heterogeneity (i.e. the differences in meaning of concepts) is, in current solutions, reduced to syntactic rewriting without explicating the meaning of the data.
When broadening the scope of the information integration problem to cross organizational border, the problem becomes even more severe, because within an organization there is a certain amount of control over the applications, while there typically is no control over the applications used by other organizations or even organizational units. Standards are arising to mitigate these problems of inter-operability between organizations, but current approaches are very specific for certain branches and are typically very rigid in what they require from anyone using the standard.
We investigate these problems of semantic information integration and their solutions using ontology technology in the context of two major European research projects.
In the COG (Corporate Ontology Grid) project we try to provide a solution for the information integration problem within enterprises. We used the Unicorn Workbench tool to create an Information Model (an ontology) with a well-defined meaning and used the Workbench to map data schemas taken from the automotive industry to the Information Model in order to make the meaning of the concepts in the data schemas explicit.
In the context of the Esperonto project, we take a broader approach and try to solve the problem of Information Integration between different enterprises. We do this in the context of the Semantic Web, where we assume the enterprises are entities on the Semantic Web that have formal explicit models for their data in the form of ontologies. We introduce a solution to the information integration problem, through the creation of explicit mappings between ontologies, in a semi-automatic manner.
Semantic Web technology, and especially ontology technology, enables information integration on a semantic level both within and across organizational boundaries, but there are still many issues to resolve, such as standardization in languages for the specification of mapping rules and the development of usable, industrial-strength tools for the creation and operationalization of these mappings.