Complex Schema Mapping and Linking Data: Beyond Binary Predicates

Currently, datasets in the Linked Open Data (LOD) cloud are mostly connected by properties such as owl:sameAs, rdfs:subClassOf, or owl:equivalentProperty. These properties either link pairs of entities that are equivalent or express some other binary relationship such as subsumption. In many cases, however, this is not sufficient to link all types of equivalent knowledge. Often, a relationship exists between an entity in one dataset and what is represented by a complex pattern in another, or between two complex patterns. In this paper, we present a method for linking datasets that is expressive enough to support these cases. It consists of integration rules between arbitrary datasets and a mediated schema. We also present and evaluate a method to create these integration rules automatically.