- From: Dan Brickley <danbri@w3.org>
- Date: Thu, 23 Sep 2004 07:26:50 -0400
- To: Dave Reynolds <der@hplb.hpl.hp.com>
- Cc: public-esw@w3.org
Dave, other SWEDers, One more SWED question. I'd like, per rough notes in http://rdfig.xmlhack.com/2004/09/23/2004-09-23.html#1095929014.190278 to be able to merge the DOAP and SWED descriptions of the FOAF project. It isn't really an opensource software project, nor a classic Environmental project, but it has aspects that can be partially described with these two sets of conventions. http://www.foaf-project.org/doap.rdf http://www.swed.org.uk/swed_data_entry/rdfit/TheFOAFProject.rdf I'd like to be able to meet scenarios such as: 'find the doap:mailing list of all things that have a swed:has_topic of swed_toi:information_resource_discovery'. ...which would require either consensus on project URIs, or merging based on OWL semantics, most likely InverseFunctionalProperty. FOAF and DOAP have foaf:homepage and doap:homepage (which are close enough to be linked using rdfs:subPropertyOf). SWED uses a string representation of the homepage (or 'primary URL', which I'm taking to be much the same thing). <swed:has_primary_url>http://www.foaf-project.org/</swed:has_primary_url> This takes us into the realm of N3 (and grungy areas of it, such as representing URIs in string form), so I fear it may currently be hard for an RDFS and OWL-subset based SemWeb crawler to do the data merging needed to service my query. Would you consider changing things, eg. by also emitting a foaf:homepage or swed:homepage property that is resource-valued? (if the latter, relating it to foaf:homepage would also be appreciated, when you do the schema). Another thing to consider is that because you've used plain literals, they'll only compare identical when they have the same xml:lang, which might also be an obstacle to loosly-coupled data merging. This bites us in FOAF slightly with the mbox_sha1sum property, btw. In that case I can live with it, since putting mbox hashes into URI space seems odd. But for things that are defined to be URIs, a more RDF-friendly idiom would be to represent them as such within the graph. I like btw the ability to go back and revise the SWED file via Web forms, and suspect that having simple string values might be associated with that... Thanks for any thoughts, Dan
Received on Thursday, 23 September 2004 11:26:50 UTC