- From: Pat Hayes <phayes@ihmc.us>
- Date: Sat, 15 Jun 2013 14:38:46 -0500
- To: Antoine Zimmermann <antoine.zimmermann@emse.fr>
- Cc: Peter Patel-Schneider <pfpschneider@gmail.com>, RDF WG <public-rdf-wg@w3.org>
Antoine, Further to my long email in response to yours, another way to proceed occurs to me. We leave the truth-conditions on graphs alone (flawed though they are, changing these now is intellectual fracking and just as dangerous) but we define entailment by a set of graphs as follows. First. we observe that when graphs share a bnode, then what they assert together is not fully expressed by considering them in isolation. Motivated by this, we define the "compounding" of a set of graphs to be the set obtained by replacing any subsets of the set which share a bnode by their union. In other words, it puts any split-up RDF molecules back together. Then we say that S entails G just when any interpretation which satisfies the compunding of S, satisfies G. In the case of no shared bnodes this is completely conventional, but it treats graphs sharing a bnode as parts of a larger graph, which is exactly how they were intended to be understood. And it makes the validity work under the mapping to FOL, provided you really do map to FOL accurately in the shared-bnode case. Would that work better for you? Peter, any comments? Anyone else, any comments? Pat PS. I am negotiable on "compounding". Precipitation? Agglutination? Restoration? Blending? Mashing? ------------------------------------------------------------ IHMC (850)434 8903 or (650)494 3973 40 South Alcaniz St. (850)202 4416 office Pensacola (850)202 4440 fax FL 32502 (850)291 0667 mobile phayesAT-SIGNihmc.us http://www.ihmc.us/users/phayes
Received on Saturday, 15 June 2013 19:39:18 UTC