The case for holistic data integration

Abstract

Current data integration approaches are mostly limited to few data sources, partly due to the use of binary match approaches between pairs of sources. We thus advocate for the development of more holistic, clustering-based data integration approaches that scale to many data sources. We outline different use cases and provide an overview of initial approaches for holistic schema/ontology integration and entity clustering. The discussion also considers open data repositories and so-called knowledge graphs.