- From: Richard Cyganiak <richard@cyganiak.de>
- Date: Mon, 21 May 2012 11:08:28 +0100
- To: Pat Hayes <phayes@ihmc.us>
- Cc: RDF Working Group WG <public-rdf-wg@w3.org>
Hi Pat, On 20 May 2012, at 03:59, Pat Hayes wrote: >> a conforming parser MUST generate an RDF graph that is isomorphic to the test case graph, except that literals MAY be replaced by other literals of equal value. >> >> (In a perfect world, RDF Semantics would define something like LV-Entailment (literal value entailment) that captures this notion. So one could say that an HTML5+RDFa parser MUST generate a graph that is LV-equivalent to the test case graph.) > > We could easily define this. It would be based on a version of D-entailment but added to simple entailment rather than RDFS-entailment, and then equivalence is simply entails-and-entailed-by, in the usual way. This sounds promising. > Of course what you get would depend upon which D you were assuming. Right. Different people/specs/implementations will assume different D and that's fine. > Actually it would probably be better to define an equivalence relation on literals and then define graph-isomorphism under this equivalence. That completely avoids any strange equivalences arising from idiosyncracies of D-entailment for particular Ds. But yes, its easy to do. Right, this would work too. I have to say I like the idea of having this defined as a form of entailment. It seems that LV-entailment would be a form of entailment that is fairly easy to grasp, and it's easy to see why it is useful; and understanding this is is a good stepping stone towards understanding the more complex forms of entailment. Also, since we already have the formalism of “entailment”, including useful additional terminology defined in terms of “entailment” such as “equivalent” or “valid”, I think that it's a good idea to use the device of “entailment” to explain relationships between graphs when possible. But then I don't understand the details that would be involved in making this all work, so if it's not feasible to define this as a form of entailment, then an extended notion of graph isomorphism will do too. > If you really think this is worth doing, by all means make it an action item on the Semantics editors. I'll raise it as an issue for now, given that it would probably require a WG resolution at some point. I'll attach an action on you to work out a good way of defining it. Best, Richard >> >> Pat >> >>> >>> Best, >>> Richard >>> >>> >>> >>>> >>>> - Steve >>>> >>>> -- >>>> Steve Harris, CTO >>>> Garlik, a part of Experian >>>> 1-3 Halford Road, Richmond, TW10 6AW, UK >>>> +44 20 8439 8203 http://www.garlik.com/ >>>> Registered in England and Wales 653331 VAT # 887 1335 93 >>>> Registered office: Landmark House, Experian Way, Nottingham, Notts, NG80 1ZZ >>>> >>>> >>> >>> >>> >> >> ------------------------------------------------------------ >> 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 >> >> >> >> >> >> >> > > ------------------------------------------------------------ > 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 Monday, 21 May 2012 10:09:03 UTC