Intelligent Support for Solving Classification Differences in Statistical Information Integration

Integration of heterogeneous statistics is essential for political decision making on all levels. Like in intelligent information integration in general, the problem is to combine information from different autonomous sources, using different ontologies. However, in statistical information integration specific problems arise. This paper is focussed on the problem of differences in classification between sources and goal statistics. Comparison with existing information integration techniques leads to the conclusion that existing techniques can only be used if individual data underlying the statistics is accessible. This requirement is usually not met, due to protection of privacy and commercial interests. In this paper a formal approach and software tools are presented to support statistical information integration, based on a generic ontology for descriptive statistics, and heuristics that work independent of the domain of application. The heuristics were acquired from economic experts working in the field of European Common Fisheries Policy.