Using semantic constraints for data verification in an open world

Abstract

For the purpose of verifying given data sets with respect to semantic criteria, OWL-based ontologies from the problem domain can be combined with verification rules expressed in the Semantic Web Rule Language (SWRL). Whenever a verification fails the user needs support for identifying the SWRL rule atoms leading to the observed failure. We present an approach that derives so-called Semantic Constraints (SCs) from the verification rules in order to spot the errors within the given data set. This procedure has the advantage of continuously retaining the Open World Assumption (OWA) within the entire modeling process. SCs are derived from SWRL rules by a stepwise reduction of the rules' content. Though our approach is generic, we draw our examples from the domain of railway infrastructure planning. © 2012 Springer-Verlag.