- From: Frank Manola <fmanola@acm.org>
- Date: Wed, 22 Oct 2003 12:48:30 -0400
- To: "Hammond, Tony (ELSLON)" <T.Hammond@elsevier.com>
- Cc: 'Brian McBride' <bwm@hplb.hpl.hp.com>, www-rdf-interest@w3.org
Hammond, Tony (ELSLON) wrote: > Hi Frank: > > >>Finally, there was a question about "whether the Primer actually cares >>if the applications it discusses are generally useful". I'm not >>entirely sure how to interpret this comment. Part of the process for >>deciding what went in there was based simply on who was willing to >>contribute. More generally, it was hoped that the applications (which >>are often rather domain-specific) that were described would suggest >>ideas to people about how they might use RDF in other domains. >> > > I should clarify that I did not in any way intend "useful" to refer to one > schema being better than another, just that there be a schema, so that the > RDF classes and properties could be married to a URI architecture (schema or > ontology, if you will), whether RDFS or OWL - otherwise just looks like > plain XML to me. (Except of course that the XML elements can then be > variously differentiated by category - class and property - based on the RDF > data model.) Sorry for any confusion I may have caused. > Tony-- Thanks for the clarification. But I think I'd also take issue with what seems to be your assumption that for an application to be "useful" it must have an RDFS (or OWL) schema. People who use RDF in their applications certainly need to understand what they mean by the terms they use (for the subjects, predicates, and objects) in the triples. However, I can imagine "useful" RDF applications not having RDFS or OWL schemas defining these things. Suppose, for example, that the source of the RDF is a database table for which there is row and column metadata already, and the target is another similar table; the RDF is just the exchange mechanism, and the definitions, while in existence, aren't redefined in RDF. Or suppose that the "schema" information is otherwise built into the source and target processing software (based, say, on UML conceptual models used to describe, or even generate, the software). The point isn't that such applications might not be *more* "useful" with RDFS or OWL schemas; the point is that the lack of such schemas does not automatically render such applications *not* useful (at least, in my opinion, and presumably in the opinion of anyone designing and using such applications). --Frank
Received on Wednesday, 22 October 2003 12:21:48 UTC