I'll pile on with a with a couple of other things: 1. I second Ed's point about conneg: http://infomotions.com/etexts/literature/english/1500-1599/more-utopia-221 should probably return a 300 code with pointers to your various file types. 2. Replace dc with dcterms (http://purl.org/dc/terms/) 3. While Ed's point about linking to other resources would be nice, first I'd focus on the resources you have and can control. Rather than a literal for dc:creator, can you mint URIs for all of your authors? How about subjects? 4. Your URIs in your rdf:Description[@rdf:about] aren't terribly helpful on their own. Either give the full URI here or add an xml:base="http://infomotions.com/etexts/literature/english/1500-1599/" attribute to the tag -- that should improve things. 5. I think your dc:contributor tag might be running aground of httpRange-14 -- I'm pretty sure you didn't help Thomas More write his story. This, I think, is the absolute hardest thing to get right with RDF/LOD. A nice example of sidestepping this sort of collision is Toby Inkster's RDF-ification of Amazon Web Services: http://purl.org/NET/book/isbn/0140449108#book -- in this example, the 'record metadata' lives at the base URI (http://purl.org/NET/book/isbn/0140449108) and the real world object lives at http://purl.org/NET/book/isbn/0140449108#book. This way Toby can claim responsibility for making the data the available, but not assert that he had any part in creating the work itself. The two resources are linked to each other, but are each unique, independent URIs. If you do do this, though, it messes up what I said in point #1. The concordances would also be really neat to see -- building off of WordNet would be pretty cool with all of these old texts. Good luck, it's great to see. -Ross. On Mon, Jul 20, 2009 at 10:04 PM, Ed Summers<[log in to unmask]> wrote: > Heya Eric: > > The main thing you'd want to do would be to make sure URIs like: > > http://infomotions.com/etexts/literature/english/1500-1599/more-utopia-221 > > returned something useful for both people and machine agents. The > nitty gritty details of how to do this can roughly be found in the > Cool URIs for the Semantic Web [1], or How to Publish Linked Data [2]. > A slight variation would be to use something like RDFa [3] to embed > metadata in your HTML docs, or GRDDL [4] to provide a stylesheet to > transform some HTML to RDF. > > The end goal of linked data, is to provide contextual links from your > stuff to other resources on the web, aka timbl's rule #4: > > Include links to other URIs. so that they can discover more things. [3] > > So for example you might want to assert that: > > <http://infomotions.com/etexts/literature/english/1500-1599/more-utopia-221> > owl:sameAs <http://dbpedia.org/page/Utopia_(book) . > > or: > > <http://infomotions.com/etexts/literature/english/1500-1599/more-utopia-221> > dcterms:creator <http://dbpedia.org/resource/Thomas_More> . > > It's when you link out to other resources on the web that things get > interesting, more useful, and potentially more messy :-) For example > instead of owl:sameAs perhaps an assertion using FRBR or RDA would be > more appropriate. > > Thanks for asking the question. The public-lod list [4] at the w3c is > also a really friendly/helpful group of people making data sets > available as linked-data. > > //Ed > > [1] http://www.w3.org/TR/cooluris/ > [2] http://www4.wiwiss.fu-berlin.de/bizer/pub/LinkedDataTutorial/ > [3] http://www.w3.org/TR/xhtml-rdfa-primer/ > [4] http://www.w3.org/TR/grddl-primer/ > [5] http://www.w3.org/DesignIssues/LinkedData.html > [6] http://lists.w3.org/Archives/Public/public-lod/ >