W3C home > Mailing lists > Public > public-owl-dev@w3.org > April to June 2019

Mapping OWL to RDF*

From: Chris Mungall <cjmungall@lbl.gov>
Date: Sat, 25 May 2019 13:57:51 -0700
Message-ID: <CAN9Aifvupv-3aSFOXGxHnj=sKhYtV4fczkQd2mm6Hb9a5nwTLg@mail.gmail.com>
To: Owl Dev <public-owl-dev@w3.org>
RDF* provides a way of unifying RDF and Property Graphs (neo4j, "knowledge
graphs", etc):


In my experience everyone ends up with similar patterns for storing TBox
axiom patterns like SubClassOf-SomeValuesFrom as single edges in PGs, and
often as single triples in triplestores (e.g. Wikidata treatment of TBoxes).

There's an obvious way to better formalize this, where the edge properties
determine the OWL interpretation. In RDF* this could be written something

<<:nucleus :part-of :cell>> owlstar:hasInterpretation
owlstar:SubClassOfSomeValuesFrom .

<<:hand :has-part :finger>>
   owlstar:hasInterpretation owlstar:SubClassOfQCR ;
   owlstar:hasCardinality 5 .

Hopefully this gives the general idea, which should be fairly obvious.

Are there any attempts to coalesce either a standard or at least a core
vocabulary for this? Would this be the right mailing list for organizing
this? Or is this already happening somewhere else?

This could be part of a larger effort to support a range of different
semantics for RDF* and PGs.

E.g. probabilistic

<<:RhinovirusInfection :has-symptom :RunnyNose>> probstar:hasProbability
0.75 .

Or contextual, a la IKL

<<:clark_kent owl:sameAs :superman>> a ikl:that ; :believed-by :lois_lane .

There are challenges in terms of layering this with existing mappings to
OWL but there are some pragmatic approaches for dealing with these.
Received on Saturday, 25 May 2019 20:58:33 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 16:41:20 UTC