Mappings in the Web of Data

The Linking Open Data (LOD) initiative has motivated numerous data providers to expose and interlink their data on the Web. But since different data providers often use different vocabularies for describing related resources, applications that consume data from the Web must first reconcile the structural and semantic heterogeneities among the data sets. As a possible solution for the representation, execution, and maintenance of mappings in the context of LOD, we propose a Web-based technique for establishing metadata interoperability. It allows for the declaration of mapping relationships between incompatible vocabularies and provides mapping features that go beyond those of existing Semantic Web languages. Further, it allows us to compile mapping specifications into executable SPARQL templates for integrating metadata from heterogeneous endpoints at run-time. With a reference implementation and a case study, we demonstrate the applicability of our proposed solution.