- From: Seth Russell <seth@robustai.net>
- Date: Thu, 16 Nov 2000 16:45:54 -0800
- To: ssarkar@ayushnet.com
- CC: www-rdf-interest@w3.org
ssarkar@ayushnet.com wrote: > Since RDF is declarative, it is kind of > out-of-synch with the application in the background. Could you clarify that - what does 'declarative' mean such that it could be out of sync with a real application? > Idea of atomicity is in transactions. One can rollback > atomic transactions as indivisible units. It seems > to me that 'Unreification' needs similar notion > of indivisible collection of sentences. Ok, i see, if a reified statement must be a 4 statement set, then all four statements must be forgotten together along with whatever else has been asserted on that node. > I think that each sentence to have an 'ident' is > a nice idea for declarative syntax. And then > reification itself is expressed by using 'ident' of > sentences and by expressing in a relationship > (a triple or quadruple etc.). > Then the problem mentioned above will not arise. Yes, I think, if we can use an ident to uniquely identify a statement and use that ident as an object, then a lot of our troubles will go away. But I doubt that the powers that be will consider sanctifying such a bold departure from M&S. However, in passing I note that, about half of the implementations of RDF listed at [1] have chosen to assign an ID to each statement. [1] http://WWW-DB.Stanford.EDU/~melnik/rdf/db.html > A relationship over sentences should also be a > sentence with an 'ident'.. and so on. > This can make semantics more explicit to internal > applications for serializations. Yes, but how do we output this in official RDF ? Seth Russell
Received on Thursday, 16 November 2000 19:44:16 UTC