- From: Pat Hayes <phayes@ihmc.us>
- Date: Fri, 16 Dec 2011 11:45:42 -0600
- To: Antoine Zimmermann <antoine.zimmermann@emse.fr>
- Cc: David Wood <david@3roundstones.com>, Guus Schreiber <guus.schreiber@vu.nl>, RDF WG <public-rdf-wg@w3.org>, Tim Berners-Lee <timbl@w3.org>
On Dec 16, 2011, at 10:29 AM, Antoine Zimmermann wrote: > Le 14/12/2011 22:09, Pat Hayes a écrit : >> >> On Dec 14, 2011, at 9:57 AM, Antoine Zimmermann wrote: >> >>> The idea of having an agreed formal semantics is to be sure that no >>> matter where and when the graph is, what you can conclude from it >>> is the same. With a semantics based on URI resolution, it becomes >>> dependent on who, where and when a user agent is when trying to >>> make conclusions. You go offline and the graph changes its meaning. >>> You're behind a proxy or haven't enough rights and you change the >>> meaning... >>> >>> We can propose this as a best practice (the graph URI /SHOULD/ be >>> used to denote the RESTful etc...) but hardly a rule in the formal >>> semantics. >> >> No, really, we can make a formal semantics which takes into account >> URI resolution. Of course it will refer to whatever is resolved, so >> will be indeterminate in that sense, but it can still be enough to >> usefully (and nontrivially) constrain the possible meanings. We have >> done this for owl:imports, for example. One can rephrase >> http-range-14 in formal semantic terms, with a great increase in >> clarity, I might add. > > The mechanism of owl:imports is not part of the model theory of OWL. I know. But it could have been. The necessary equations and supporting definitions are part of the ISO Common Logic standard. > With the formal semantics of OWL, a given RDF graph has a unique set of possible entailments, no matter what's the state of the network. > owl:imports serves to define what *is* the RDF graph on which to define this unique set of entailments. It acts prior to the definition of interpretation, so to say. That is one way to do it, but it could also have been done semantically. Or indeed both, since the point of the semantic account is to support the 'real-life' interpretation in cases like this. > In terms of what owl:imports /denotes/, there are no constraints other than the domain and range are both owl:Ontology. I could very well interpret owl:imports as "this ontology is funnier than that ontology", and it would be perfectly consistent with the semantics. Which is a weakness of the current OWL semantics. And which would, by the way, break applications which use owl:imports, right? > > Similarly, we could very well define a mechanism which says that the RDF graph associated with a URI is defined as the set of triples that you can get from a HTTP lookup. Then the set of triples is known and fixed and you use the normal RDF semantics on this, which gives the same entailment for everybody, everywhere, always, and which can interpret the URI you looked up as just anything. True, but we don't have to impose the universal, known, fixed part. For example, the truth condition for owl:imports is owl:imports UUU is true in I just when I(I(UUU)) = true which uses the interpretation mapping itself to find the denotation of the IRI, then interprets *that* to get the truthvalue. Thus, as one would expect, the truthvalue depends on the interpretation of that naming IRI just as much as it does on any other IRI. How "fixed" this is depends on what we say about I(UUU) in cases like this. The named-graph semantics used in our old paper has the property that any interpretation satisfying the naming constructions has to have I(UUU) be the named graph itself; but one could also just say that I(UUU) is determined by the state of the Web, and refer to http-range-14. But, whatever graph it is, the meaning of importing it is just like that of asserting it where it was imported, is the point (and is what that semantic equations says.) Note, asserting it, not copying it. > > In fact, this would already be a good step forward, as in RDF 2004, nothing says what delimits a graph. With RDF 2004, it is not clear that all triples in a single file necessarily belong to the same graph. ? They can belong to many graphs, of course. I agree, RDF 2004 is vague on graph boundaries: that is exactly why I suggested introducing a notion of scoping in my ISWC talk http://www.slideshare.net/PatHayes/blogic-iswc-2009-invited-talk (see slides 15-25) > Nor is it clear that triples in different files or at different places belong to different graphs. In RDF 2004, we can very well believe that there is just one giant graph composed of all triples ever published online. Indeed. Is this a problem? > > With a semantics like Tim proposes (although I would not like it to be included in the model-theoretic semantics), we would make clear what's in a graph in function of what you get in a lookup, just like OWL makes clear what's in an ontology thanks to expliciting the (non-model-theoretic) semantics of owl:imports. > > Anyway, back to the discussion, David said that Tim says "the URI denotes", and I interpret this word as "what the model-theoretic interpretation assigns to the URI". Me too. Pat > > > AZ > >> >> If I might add a meta-comment, it would be useful if the members of >> the WG were to say what they want the semantics to do, and let the >> editors tell them what is or is not possible. >> >> Pat >> >>> >>> Le 14/12/2011 00:29, David Wood a écrit : >>>> Hi all, >>>> >>>> I had a lengthy conversation with TimBL about named graphs at the >>>> LEDP Workshop [1] last week. Briefly, he feels that the >>>> semantics for named graphs should work like this: >>>> >>>> - An RDF Graph is named via a URI. - The URI denotes the RESTful >>>> Representation that is returned when the URI is resolved. >>>> >>>> That is, the URI denotes the graph's contents, not the graph >>>> Resource itself. >>>> >>>> How do Peter and Pat feel about that? >>>> >>>> TimBL: Please let us know if I misrepresented your position. >>>> >>>> Separately, Elsevier representatives Brad Allen and Alan Yagoda >>>> informed me that by "named graphs" they mean an RDF Graph that is >>>> referenced by a URI. Resolution of that URI returns the graph >>>> contents (a g-text) via RESTful interaction. That would seem to >>>> be in line with TimBL's preference. >>>> >>>> Regards, Dave >>>> >>>> >>>> >>>> >>>> On Dec 13, 2011, at 15:54, Guus Schreiber wrote: >>>> >>>>> All, >>>>> >>>>> It is quiet on the mailing list. The main thing we seem to be >>>>> in limbo about is the GRAPHS debate. I suggest we devote the >>>>> meeting to this theme. I have included in the agenda some >>>>> discussion topics that came up in recent telecons, plus the >>>>> email of Andy on TriG examples. I suggest we also have a >>>>> meta-discussion on what our options are for getting consensus. >>>>> >>>>> The agenda is at: >>>>> >>>>> http://www.w3.org/2011/rdf-wg/wiki/Meetings:Telecon2011.12.14 >>>>> >>>>> Hope to speak to many of you tomorrow. Guus >>>>> >>>> >>>> >>>> >>> >>> -- Antoine Zimmermann ISCOD / LSTI - Institut Henri Fayol École >>> Nationale Supérieure des Mines de Saint-Étienne 158 cours Fauriel >>> 42023 Saint-Étienne Cedex 2 France Tél:+33(0)4 77 42 66 03 >>> Fax:+33(0)4 77 42 66 66 http://zimmer.aprilfoolsreview.com/ >>> >>> >> >> ------------------------------------------------------------ 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 >> >> >> >> >> >> > > -- > Antoine Zimmermann > ISCOD / LSTI - Institut Henri Fayol > École Nationale Supérieure des Mines de Saint-Étienne > 158 cours Fauriel > 42023 Saint-Étienne Cedex 2 > France > Tél:+33(0)4 77 42 83 36 > Fax:+33(0)4 77 42 66 66 > http://zimmer.aprilfoolsreview.com/ > > ------------------------------------------------------------ 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 Friday, 16 December 2011 17:46:20 UTC