A proposal for extending RIF RDF and OWL compatibility towards OWL 2

I first summarize my understanding of OWL 2 before formulating my
proposal for extending the RIF RDF and OWL compatibility document [0] to
OWL 2 [3,4].

I think we should discuss this proposal in the RIF+OWL coordination
telecon next week.

I cannot post to the public-owl-wg mailing list, so I included everyone
from that group who answered Sandro's poll [7].  If you believe this
e-mail is of interest to anyone not in the 'To' of the email, please
feel free to forward.

==
OWL 2 Full [4]
==

- is syntactically the same as OWL 1 Full, i.e., the set of OWL 2 Full
ontologies is the same as the set of RDF graphs
- is semantically in some parts an extension and in other parts a
restriction of the OWL 1 Full semantics, the main restriction being that
there are no comprehension conditions

==
OWL 2 DL [3,5]
==

- The syntax and semantics of the language is defined in terms of a
functional style syntax that is not comparable with the abstract syntax
of OWL 1 DL.  This syntax corresponds to a structural model.  There
exist various serializations of this structural model for exchange
purposes  (e.g., XML, RDF).
- the semantics specification is different from, but similar in spirit
to the semantics of OWL 1 DL, the main difference being the fact that
annotation properties are no longer interpreted.
- the RDF syntax is slightly different, but it is my understanding that
given two OWL 1 DL ontologies O1 and O2 such that O2 does not contain
ontology or annotation properties, and their RDF graph forms R1 and R2,
the mapping from RDF graphs to OWL 2 ontologies in [1] is such that if
the mapping is applied to R1 and R2, the resulting O1' and O2' are OWL 2
DL ontologies and O1 owl-1-dl-entails O2 iff O1' owl-2-dl-entails O2'.
So, OWL 1 DL ontologies in RDF graph form that do not contain annotation
properties can be seen as OWL 2 DL ontologies.


==
OWL 2 RL [6]
==

- OWL 2 RL is a syntactic subset of OWL 2 DL with the same semantics
- there exists a set of forward chaining rules that can be used to
implement certain OWL 2 RL inferences on a subset of the OWL 2 RL
ontologies in RDF graph form.  These inferences are class membership,
property value, and equality and inequality statements. These rules can
be straightforwardly written as RIF rules [2].



================================
Extending RIF RDF and OWL compatibility
================================

There are 4 separate issues to be considered:
- specification of OWL 2 Full compatibility
- specification of OWL 2 DL compatibility
- specification of import profiles (section 5)
- embedding of RIF-OWL 2 RL combinations into RIF-BLD (section 8.2)


=========
Specification of OWL 2 Full compatibility
=========
This specification is completely straightforward.  It is a very simple
adaptation of the specification of OWL Full compatibility.
(P1) I propose that this specification replaces OWL Full compatibility,
because I believe that for OWL Full users the fact that there are no
comprehension conditions is not the real limitation (and actually cause
problems in the specification).


=========
Specification of OWL 2 DL compatibility
=========
This specification should be a relatively straightforward adaptation of
the current specification of OWL DL compatibility.
One could also argue that this specification should replace the current
specification of OWL DL compatibility because of the mentioned
correspondence between OWL 1 DL ontologies in RDF graph form and OWL 2
DL ontologies.  However, in that case it is not possible to consider
annotations in entailed ontologies.  This might not be a limitation,
though, since current implementations of OWL DL do not consider
annotations in the semantics anyway.
(P2) I thus propose to replace OWL DL compatibility with OWL 2 DL
compatibility.
If OWL DL compatibility is replaced with OWL 2 DL compatibility, it
seems awkward to retain OWL DL-annotation compatibility, since it can
only be used with OWL 1 DL.  Therefore:
(P3) I propose to drop OWL DL-annotation compatibility (section 4.2.2.3).

=========
Profiles for imports
=========
(P4) I propose to leave the IRIs of the profiles as they are currently
specified in section 5.1.1, since the combinations still work for OWL 1.

The question is then which syntaxes to allow for imports.  Currently,
the only allowed syntax is RDF.
I could imagine that we would also allow XML, to cater for the XML
syntax of OWL 2, but I don't have a strong opinion here.


========
Embedding of RIF-OWL 2 RL combinations into RIF-BLD
========
If we adopt proposal (P2), we need to update section 8.2 as well.
(P5) I propose to update the embedding in section 8.2 to RIF-OWL 2 RL
combinations.
It of course remains to be seen whether we can embed all RIF-OWL 2 RL
combinations, but I suspect this is possible.

I guess it is important to point out the difference between this
embedding and the embedding of the above-mentioned RDF-based entailment
rules [2] (both in [0] and in [2]).


I believe these are all the substantive things to be changed.  Obviously
it means we need to reissue last call.
If we decide quickly (i.e., next week) about my proposals (P1-P5) I can
finish the update of the document (save the proof of faithfulness of the
RIF-OWL2RL combinations embedding) this year (2008), and I think it
should be possible to publish the document as last call towards the end
of January 2009.


Best, Jos


[0] http://www.w3.org/2005/rules/wiki/SWC
[1] http://www.w3.org/TR/2008/WD-owl2-mapping-to-rdf-20081202/
[2] http://www.w3.org/2005/rules/wiki/OWLRL
[3] http://www.w3.org/TR/2008/WD-owl2-semantics-20081202/
[4] http://www.w3.org/TR/2008/WD-owl2-rdf-based-semantics-20081202/
[5] http://www.w3.org/TR/2008/WD-owl2-syntax-20081202/
[6] http://www.w3.org/TR/2008/WD-owl2-profiles-20081202/
[7] http://lists.w3.org/Archives/Public/public-owl-wg/2008Dec/0020.html

-- 
Jos de Bruijn            debruijn@inf.unibz.it
+390471016224         http://www.debruijn.net/
----------------------------------------------
No one who cannot rejoice in the discovery of
his own mistakes deserves to be called a
scholar.
  - Donald Foster