It's also worth noting that labels can be anything. If you don't expect people ever to see the labels, you can make them whatever you want (label: "Thing 1835"). This is an even easier perl script :)
A Bnode actually is one that doesn't have a URI, not one that doesn't have a label. I think nodes without URIs are a terrible idea. On the other hand nodes without labels is not such a big problem (for RDF). Also, one could certainly modfy exhibit to mint random labels for items that are not given one. This might be more convenient at times, but it would also surely happen that those gibberish names would sometimes be exposed accidentally. So arguably forcing labels ensures some degree of niceness in the exhibits that created. -David David Huynh wrote: > You're right that the labels here are redundant. If you want to model > your data in pure RDF, then you would use "bnodes" for these records, > which are essentially tertiary relationships (product, location, price) > rather than just binary relationships. Exhibit does not have a notion of > "bnodes" and it requires you to provide a label for everything. > > Exhibit can be made to understand the notion of bnodes. But there is not > a strong enough demand for that, and besides, within the RDF community, > I believe that bnodes are still considered both a blessing and a curse. > So I'm going to hold off while the jury is still out on that one. > > By the way, if you're so determined to choose the "best" RDF format, you > might miss out on opportunities to pick the formats that work for > particular situations. Remember, the data model (graph with URIs) is > much more important than the serialization format. > > But if you want to hold steadfast to something blessed like RDF/XML, > that's fine too. This example > http://people.csail.mit.edu/dfhuynh/projects/data-theft/data-theft.html > shows how Exhibit can load an RDF/XML file. Actually, Exhibit > automatically routes the file through Babel for conversion. > > David > > Rahul Nabar wrote: > >> Thanks David and Alexy, >> >> I think both those options work for me. But I had a question about the >> data structure you suggested David: >> >> { label: "Tea in London", >> type: "Pricing", >> city: "London", >> product: "Tea", >> price: 5 >> }, >> >> >> Isn't there a redundancy in label choice here? I mean I could probably >> use a perl / php preprocessing script to produce the "label" fields. >> >> But the Label as we would like it here is only a derived epithet. i.e. >> a combination of the "product" and "city" fields. >> >> Is there a way to get Exhibit to generate them on the fly? Just goes >> down to my concern of choosing the "best" RDF format. >> >> Thanks! >> >> -Rahul >> ------------------------------------------------------------------------ >> >> _______________________________________________ >> General mailing list >> [email protected] >> http://simile.mit.edu/mailman/listinfo/general >> >> > > _______________________________________________ > General mailing list > [email protected] > http://simile.mit.edu/mailman/listinfo/general > _______________________________________________ General mailing list [email protected] http://simile.mit.edu/mailman/listinfo/general
