rdfweblog example, 2000-09-09

danbri's semi-fictional weblog for Sept 9th, 2000

(disclaimer: ok, this is dull. I don't do weblogs! this is just some example text to play with + explanation... Do a view/src to see the semantic annotations I'm playing with to allow this sort of narrative to be summarised for Semantic Web apps...)

Today I have been reading the book Cryptonomicon much of the day, to take my mind off the RDF Issues List. Then I played for a brain-rotting while on my Dreamcast console. Then I watched the The Mummy and read Cryptonomicon while Martin, Craig and Liz watched Air Force One, since the film was preposterous honk. Now I'm listening to some music (actually David Bowie's Low, whose tracklistChecksum I don't have to hand) and chatting in IRC with Jonathan Eisenzopf when I should be getting some sleep...

We were talking about variations on the weblog idea, the relationship between that and XHTML 'semantic web screenscraping', and I wanted to show the weblog extension to the rdfwebring system that I've been designing. Being too verbose for IRC, I wrote it up as this example file.

So... here's a sketch of an RDFWebLog data file. The yntax is XML tagged with significant verbs and named items, which are referenced indirectly, eg 'the person whose homepage is...' or 'the film reviewed at...', just as we do in the RDFWebRing.

This example sketch is only intended to be evocative of what we might build with RDF; there are plenty of bugs, and the WordNet representation needs further thought. Basic idea: we can mention objects through uniquely-identifying description in our WebLogs. Postprocessing of this into HTML or into richer RDF representations (eg. based around Event models such as the Harmony Project's ABC) should be easy...

This example is a strange mix of XHTML and (fictional) rdfweblog markup. While it seems to render OK in HTML browsers, the objects that I mention (films etc) don't get hyperlinked. An XSLT transformation into HTML that used A HREF would do the job, though I suspect some jiggling about with the way I use XML here might allow a single format to function both as hypertext XHTML and as something that permits a rich RDF view of the data. That's the theory anyway...