- From: Graham Klyne <GK-lists@ninebynine.org>
- Date: Thu, 21 Jan 2010 11:34:32 +0000
- To: Jeremy Carroll <jeremy@topquadrant.com>
- CC: Sandro Hawke <sandro@w3.org>, David Booth <david@dbooth.org>, Harry Halpin <hhalpin@ibiblio.org>, semantic-web@w3.org
FWIW, I have sympathy both ways. I didn't think Sandro was actually suggesting changes to the RDF spec. Rather an experimental program to use interpretation properties to replicate the datatyping functionality, along the lines you suggest "define your own semantic extension which includes interpretation properties". If successful, it should be possible to map between both forms. Then we might, in time, have a basis for a proper evaluation, and coming to a clearer consensus than a 70/20 ~ 40/60 flip-flop. Maybe datatyping comes to be seen as an optimization of the other form? Or vice versa? An area in which I have a lingering background interest is semantic web ideas deployed in small embedded devices. Being able to create a simple stack in which the current datatyping mechanisms are replaced by a select few datatype properties is something I find very appealing. Being able to do so in a way that is at some level compatible with existing RDF specification and usage very much adds to that appeal. (But I think we don't need to change RDF to do this.) (Pat raised the problem of missing information - I'm not sure how that might be handled, though my intuition is that if key triples are missing then the semantics should be weakened to almost nothing.) #g -- Jeremy Carroll wrote: > > > Sandro: > "That suggests zero evolvability of RDF." > > Hmmm. > If we reopen the datatyping can of worms the vote might split 70/30 the > other way this time; and ... > then next time round it might be 40/60 back again. > > I'm trying to thing of an equivalent evolution ... there was XML DTD to > Schema .... > > With RDF 1 as specified, it is possible to define your own semantic > extension which includes interpretation properties. > I guess I would like to see success with this before adding such > interpretation properties to the spec. > But I am not quite sure what I would mean by success. > > Jeremy > > >>> I don't think 100% compatibility is enough. Adding a second mechanism >>> will simply add to confusion. >>> >>> This was considered and rejected by the earlier group, even if that >>> decision with hindsight was not brilliant (it was a 50/50 split). >>> I preferred the other design, but I have to learn to live with the >>> one we agreed. >>> >> >> That suggests zero evolvability of RDF. > >
Received on Thursday, 21 January 2010 11:37:20 UTC