RE: SSFS & Direct Semantics clarification: Invalid or inconsistent literals

>-----Original Message-----
>From: public-owl-wg-request@w3.org [mailto:public-owl-wg-request@w3.org]
>On Behalf Of Mike Smith
>Sent: Friday, June 12, 2009 6:33 PM
>To: W3C OWL Working Group
>Cc: Birte Glimm
>Subject: SSFS & Direct Semantics clarification: Invalid or inconsistent
>literals
>
>I am trying to resolve ambiguity with respect to literals for which
>the lexical value is not permitted by the datatype.  E.g.,
>
>"string"^^xsd:decimal
>"1.5"^^xsd:integer
>
>The question is whether such literals are (1) prohibited by the
>structural specification or (2) inconsistent according to the Direct
>Semantics.
>
>I find (2) to be more intuitive and I think it more closely aligns
>with the RDF Based Semantics (to which I defer to Michael Schneider).
>But...

Ok, for completeness, to show the difference to the Direct Semantics: 
In the RDF-Based Semantics, having the two triples 

  (1) :s :p "string"^^xsd:decimal .
  (2) :p rdfs:range xsd:decimal .

is syntactically valid (since it is well-formed RDF) and so has a 
semantic meaning in OWL 2 Full. But the graph is in fact semantically 
inconsistent, since the "ill-typed" literal "string"^^xsd:decimal 
denotes an individual that is not an instance of xsd:decimal 
(not even of rdfs:Literal), because "string" is not in the lexical 
space of the datatype xsd:decimal; see the 3rd "general semantic 
condition for datatypes" in Section 5.1 of the RDF Semantics [1] 
("otherwise IL("sss"^^ddd) is not in LV").

There are two interesting bits (interesting at least for those 
who care :) ):

(i) Triple (1) alone does not build an inconsistency, but simply makes 
an assertion about some individual outside rdfs:Literal. Well, 
assertions about non-data individuals are hardly special; the
interesting thing is that in this case the non-data individual 
is actually denoted by a literal.

(ii) Even if there are both such triples, this doesn't automatically mean 
that this will always lead to inconsistency. The semantic condition 
mentioned above only applies to datatypes belonging to the datatype map 
being used. For example, the above two triples clearly build an 
inconsistency in OWL 2 Full, since xsd:decimal is part of the OWL 2 datatype 
map. But xsd:decimal is /not/ a mandatorily part of the OWL /1/ datatype map,
and so the example is OWL 1 Full /consistent/. However, with every 
OWL 1 Full interpretation that uses an extended datatype map 
containing xsd:decimal, the above example graph will not be satisfied.

(Well, was it me who claimed that RDF-Based semantics is "more intuitive"
in this respect? ;-))

Cheers,
Michael

[1] <http://www.w3.org/TR/rdf-mt/#defDinterp>

--
Dipl.-Inform. Michael Schneider
Research Scientist, Dept. Information Process Engineering (IPE)
Tel  : +49-721-9654-726
Fax  : +49-721-9654-727
Email: michael.schneider@fzi.de
WWW  : http://www.fzi.de/michael.schneider
=======================================================================
FZI Forschungszentrum Informatik an der Universität Karlsruhe
Haid-und-Neu-Str. 10-14, D-76131 Karlsruhe
Tel.: +49-721-9654-0, Fax: +49-721-9654-959
Stiftung des bürgerlichen Rechts, Az 14-0563.1, RP Karlsruhe
Vorstand: Prof. Dr.-Ing. Rüdiger Dillmann, Dipl. Wi.-Ing. Michael Flor,
Prof. Dr. Dr. h.c. Wolffried Stucky, Prof. Dr. Rudi Studer
Vorsitzender des Kuratoriums: Ministerialdirigent Günther Leßnerkraus
=======================================================================