The surface (or, foreground) structure of linked data and their associated
OWL vocabularies can be complemented by background models expressing valid
ontological distinctions that may have become obscured by the modeling style
chosen by the vocabulary designer. Background models can generally serve for
debugging, visualization, matching, or even pattern-based design of
operational ontologies such as linked data vocabularies. An example of a
well-known background model language, primarily suited for taxonomic
ontologies, is the system of OntoClean meta-properties.
We present an alternative type of background model language, dubbed PURO,
which is oriented towards linked data ontologies, and relies on
particular-universal and relationship-object dichotomies. With this language
a background model (i.e., an ontologically relevant model) of each
vocabulary can be constructed and both models then can be mapped. We will
next discuss how such a model can be used to better understand the nature of
the entities of the foreground model and also how the ontological coherence
of the foreground model can be
(automatically) checked.