Six months have passed since the first release of 'The Wheel and the Hub'. Thought it was time to revisit it, and make it consistent with a parallel new release of SPEK vocabulary. The new presentation is much shorter and more full of emptiness. Eventually I've came back to the first impression about it, which is well known to often turn out to be the best one. Blank Nodes are indeed the clue for representation of subjects, but with a subtle and important difference with the first version. Blank nodes representing subjects must be really empty, they have to bear absolutely no declaration of any property whatsoever, not even links to their various descriptions (as they did in the previous versions). It's up to the various descriptions to point to the same blank node, as so many fingers pointing at the Moon, and not the other way round ...
Following this logic, SPEK vocabulary has also be simplified to the extreme ... I don't need "views" and "aspects" any more. Any RDF description is a view and provides a specific aspect. Note also that 'hubject' is back in the SPEK vocabulary, but no more as a class, but as the property linking a description to the binding blank node.
I guess this is now as simple as possible ...
My own observation is that the google page-rank algorithm reads more like "so many fingers pointing at the Moon, and not the other way around..."
ReplyDeleteJack also added in a private communication that this version is really too sparse, and would need extra prose in the form of a proper paper. Actually this is done, but in French only so far.
ReplyDelete