- From: pat hayes <phayes@ai.uwf.edu>
- Date: Sun, 20 May 2001 12:58:57 -0500
- To: "Danny Ayers" <danny@panlanka.net>
- Cc: www-rdf-logic@w3.org
- Message-Id: <v0421015eb72db494b05b@[205.160.76.183]>
> > >I also suspect that there is something happening at a psychological >level in these arguments - it's far easier for a human to relate to >information in a structure like that below - but is this necessarily >the case for machines? > > >Yes, there are significant computational advantages. In general, the >more stuff you can incorporate into the unifier, the more efficient >the inference search: you tend to be trading an exponential for a >linear cost, a very good trade. > > >I'm not really convinced, but am not going to press the point as you >are considerably more familiar with the territory than I am. >Instead, I'd like to ask you to give the case *in favour* of triples >- surely there's a baby in the bath somewhere? When just about everyone in the RDF world has bought the arguments for that case, I feel a contrarian moral duty to make the other one. The case for, as I understand it, is simply that triples are simple and they are capable of encoding any graph structure, and a lot of people seem to like them, so why not use 'em? And in fact I think that is reasonable, as long as one faces up to the fact that this uniform use of triples is going to be encoding a mixture of a number of uses that need to be distinguished from one another eventually, so the 'uniformity' can be overstated and doesnt buy one any miracle cures. In particular, it doesn't buy one a simple universal assertional framework. Pat Hayes --------------------------------------------------------------------- IHMC (850)434 8903 home 40 South Alcaniz St. (850)202 4416 office Pensacola, FL 32501 (850)202 4440 fax phayes@ai.uwf.edu http://www.coginst.uwf.edu/~phayes
Received on Sunday, 20 May 2001 13:58:52 UTC