Re: datatypes and MT

Why rdfs:str="xyz" rather than [ rdf:value "xyz"; rdf:type rdfs:Literal ]?

If we are grounding our literal data type taxonomies in RDFS, then
it seems far more symmetrical to continue with the present mechanisms:


  :s eg:shoeSize [ rdf:value "10"; rdf:type xsd:Integer ].
  :s nav:date [ rdf:value "2001-11-02"; rdf:type xsd:Date ].
  :s nav:flightNum "1154"; rdf:type xsd:Integer ].
  :s apt:elevation "58"; rdf:type unit:meter ].
  :s :p "ooo".

where in the last case, one may be inferring an implied
type which may not be specified itself in the statement.
I.e. the resultant graph would correspond to the expansion
of the above statement into:

  :s :p [ rdf:value "ooo"; rdf:type rdfs:Literal ].

In all cases, for all literals, the graph representation
could employ an anonymous node which assigns type and
which has a manditory, single value for the rdf:value property.

Consistent, generic, flexible. Allows one to assign any
number of qualifications to the literal (even such things
as the base notation of the lexical representation) and
provides a partitioning between otherwise equivalent 
literal strings which may actually represent different
"things" despite their equivalent lexical representation.

If, beyond that, one needs more explicit, global identity for
a given literal value, then they can employ URI encoding of
the literal as a URV. I.e.

  :s eg:shoeSize xsd:Integer:10 .
  :s nav:date xsd:Date:2001-11-02 .
  :s apt:elevation unit:meter:58 .

Thus, all literals recieve a manditory type in the graph or
are encoded in URVs for which type is defined in terms the URV
scheme and the URV encoded "literal" can then act as the 
subject of statements, being a proper "resource".

Eh?

Cheers,

Patrick


--
               
Patrick Stickler              Phone: +358 50 483 9453
Senior Research Scientist     Fax:   +358 7180 35409
Nokia Research Center         Email: patrick.stickler@nokia.com

Received on Monday, 5 November 2001 02:01:41 UTC