Re: RDF typed node implementation experience

At 16:25 10/10/2002 +0300, Janne Saarela wrote:

Hi Janne,

Thank you for this information.  It is certainly timely and we were unaware 
of this.

I have a couple of questions for you:

1) What does your software do with literals for which there is no datatype 
information, i.e. given:

   <rdf:Description rdf:about="A">
     <foo:prop1>10</foo:age>
   </rdf:Description>
   <rdf:Description rdf:about="B">
     <foo:prop2>10</foo:age>
   </rdf:Description>

In the absence of any range information about the two properties does your 
implementation regard the 'values' of the two properties as being 
equal?  It can be difficult to tell.  In some implementations you can and 
in some you cannot.  And in RDFCore we can argue for a long time whether 
you can tell or not :(

2) How big a burden to you would it be, if to be conformant, you had to add 
a datatype attribute to the instance data, e.g. you had to write:

   <rdf:Description rdf:about="Jenny">
     <foo:age rdf:datatype="&xsd;integer">10</foo:age>
   <rdf:Description>

Brian



>Hi all
>
>Following this week's discussion on RDF Core WG list I
>thought I share some of Profium's experience in implementing
>data type support in literal nodes. In fact, I could have
>cross-posted here a recent email I sent to www-mobile [1].
>
>Out of the several options [2] discussed by the WG,
>Profium Semantic Information Router (SIR) V3.0.4 [3]
>implements proposal F.
>
>Currently we encode data type information about
>a literal value of a property by making the 'range'
>predicate point to a URI that specifies the data type.
>For example,
>
><rdf:Property rdf:ID="price">
>   <rdfs:range rdf:resource="http://www.w3.org/2001/XMLSchema#double" />
></rdf:Property>
>
>Profium SIR does not support all of the data types defined
>by XML Schemas but only a limited subset. What remains us an
>open issue is how to make 'range' pointers to user-defined
>data types such as email addresses.
>
>Profium SIR takes advantage of 'range' typed literals
>in the following ways:
>
>1. RDF schemas are used to validate incoming native RDF instance data
>
>2. RDF schemas are used to enforce correct entry of RDF metadata
>in a Web based interface. Profium SIR uses automated JavaScript
>+ HTML creation from the RDF schema file.
>
>3. schemas are used to guide the query language Profium
>has developed. Our query language is called RDFQL even though it
>is not related to other languages by the same name.
>Just like querying of XML, RDF querying requires a parser
>to provide the application layer i.e. the query engine
>the data typing information. This information in turn enables
>us to do advanced querying on RDF instead of plain string
>matching. For example, we can look for RDF data where a date
>was before 10-Oct-2002.
>
>The implementation was straight-forward for items 1 and 2.
>The query language has certainly required more work but
>the starting point i.e. the PSVI of RDF has provided
>us a good kick-start.
>
>Regards,
>Janne
>
>[1] http://lists.w3.org/Archives/Public/www-mobile/2002Sep/0006.html
>[2] http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Oct/0049.html
>[3] http://www.profium.com/gb/products/sir.shtml
>--
>Janne Saarela <janne.saarela@profium.com>
>Profium, Lars Sonckin kaari 12, 02600 Espoo, Finland
>Tel. +358 (0)9 855 98 000 Fax. +358 (0)9 855 98 002
>Mob. +358 (0)40 508 4767  Internet: http://www.profium.com

Received on Thursday, 10 October 2002 14:21:03 UTC