- From: Pat Hayes <phayes@ihmc.us>
- Date: Thu, 13 May 2010 00:03:42 -0500
- To: Dan Connolly <connolly@w3.org>
- Cc: Jonathan Rees <jar@creativecommons.org>, AWWSW TF <public-awwsw@w3.org>
Dan, I don't think I've got my point across, and its getting lost in all this confusion about information resourceness. Its really a very simple point, and I can make it with a very simple example. Suppose A is an RDF graph, and B is an RDF/XML file which encodes/is a surface syntax of/represents (choose your favorite terminology) that graph A. And suppose U is a URI which "identifies" B, in the sense that what you get back, when you do an HTTP GET using U, is a 'representation' (in the REST sense) of B with a 200 code attached. That is, the relationship between U and B is exactly like that between the URI of a web page, and the web page itself. My point is simply that under these circumstances, we are pretty much obliged by http-range-14, as I understand it, to say that U denotes B; that is, it denotes the thing it HTTP-identifies. And if it denotes B, then it cannot denote A, since (for other reasons, on which we agree) A is not identical to B. As for being an information resource, as I say I see this as secondary and not particularly important, but I would observe that surely if anything can be an information resource, then B *certainly* is one; and that this is all that is necessary for the argument to go through. Even if A is an information resource, U cannot refer to it, since B's claim upon the reference of U is prior to, and based on a more secure case than, A's. If you want to object that there is no natural priority here, and that we humans can simply assert by fiat that the referent of U is whatever we decide it should be, in particular it could be A, I would make the following replies. First, the parallel with 'normal' URI usage in the non-semantic Web seems to me to be compelling. (In fact, surely this was the primary motivation for the http-range-14 decision in t he first place? That is, we all have to admit that the pre-semantic Web uses URIs to effectively 'name' things like Web pages, and it would be folly to adopt a SWeb convention that tried to deny this.) Second, if we decide in this example that U denotes A, how can anyone ever refer to B? With this convention, B has become semantically invisible, an unnameable entity. But B is surely as "real" as A is, in fact if anything it has more of a claim to be a real computational entity, as opposed to (what some would call) a mere abstraction. B may for example have a provenance, an author, a date of creation and even a financial value, none of which could possibly apply to a purely Platonic entity such as A. (Remember, an RDF graph is a *set*.) And third, the relationship between A and B is highly asymmetric. Given B, one can compute A. B has all the information about A. But not the reverse: A is simply a set, and has no information about how this set is syntactically encoded in a linear convention that can be put into a byte stream. I think (?) you understood me to be saying that U denotes the 'representation' of B, rather than B itself. However, that was not my intention. Part of the problem we have here is that this word, "representation" has acquired a technical meaning in this area of discussion which is rather different from its general meaning (or maybe more accurately, a very, very narrow subcase of the more general meaning.) As used in Roy's thesis and the subsequent literature, where it is the R in REST, it has the very special meaning of the relationship between an HTTP message body and the file or perhaps script that emits this body at the HTTP endpoint. I do not pretend to be able to give, or even fully understand, the full nuances of architectural finesse that are involved in stating this properly, but it certainly requires both the 'representation' and the thing 'represented' to be things in a computer memory, which have an especially close relationship to one another in that the former can completely 'capture' the latter, without loss of information. My own intuitive picture of this is simply that of a file copy, or the taking of an imprint from a platen in a hand press. But the relationship between (for example) a photograph of me and me; or between an RDF description of the familial relationships among a group of people, and the actual people; or between the XML file B and the abstract graph A, are not this kind of 'representation' at all. So it is easy to get muddled when we use this word carelessly (as I admit to doing in the previous message.) Pat On May 12, 2010, at 10:31 AM, Dan Connolly wrote: > On Tue, 2010-05-11 at 22:36 -0500, Pat Hayes wrote: >> On May 11, 2010, at 1:01 PM, Dan Connolly wrote: >> >>> On Mon, 2010-05-10 at 15:25 -0500, Pat Hayes wrote: >>>> Let me give an intuitive case in support of the Nays here. An RDF >>>> graph is a set, which is not the same as a document, for sure. The >>>> *same* graph can be encoded in a variety of different syntactic >>>> forms. >>> >>> Meanwhile, the same resource can have a variety of representations. >>> >>>> Consider two documents, one in RDF/XML, the other in NTriples, >>>> describing the same graph. If we identify the document with the >>>> graph >>>> it describes, then these have to be the same. >>> >>> But if we say that those documents _represent_ the graph, >>> they don't have to be the same. >> >> Yes, quite. The document is not the graph. No document is ever a >> graph. >>> >>>> But they aren't the >>>> same. So even if a graph is an information resource (and I agree >>>> that >>>> one can make out a case for that position), it certainly isn't the >>>> same information resource as any document (In RDF/XML or NTriples >>>> or >>>> any other notation) that represents it syntactically. >>> >>> But it can be represented by them. >> >> Indeed. But it is distinct from them, is my (only) point. > > I thought you were showing an argument that graphs are not > information resources. > >>>> So, one ought to >>>> use redirection to refer to it, according to http-range-14. >>> >>> I don't see that this follows. >> >> Well, according to http-range-14, as I understand it, if a bare URI >> gives a 200-level response to an HTTP GET, then it (the URI) denotes/ >> refers to the resource that emits that response. And in these RDF- >> graph-representations, > > yes, so far... > >> this is always the syntactic entity at the >> other 'end' of the HTTP response. > > No. The observable syntactic entity is the representation, not > the denoted resource. > >> (Because these things - the >> syntactic documents which encode the various syntactic >> representations >> of RDF graphs - are fully fledged information resources in their own >> right, so are refer-to-able by URIs, > > yes... > >> and are therefore indeed so >> referred to by them, > > I think you went too far there. Given <http://example/xyz> > if I do a GET on that, you can look at what comes back > as a document/information-resource, but you don't have > any evidence that it is called <http://example/xyz>. > You only know that it _represents_ something called > <http://example/xyz>. > > >> if they respond with a 200-coded response.) So, >> such URIs cannot refer to the graph itself (because, as we have >> already agreed, the graph is never identical to a document of any >> kind, ie to any information resource that can return a 200-coded >> response.) > > I find it pretty straightforward to deal with this stuff formally > rather than in prose. > > If we take w:representation to be the relationship between > a resource and any of its representations, then we have > this axiom in the case of HTTP: > > { > _:m a http:OKResponse; > http:about ?R; > mime:body ?BYTES; > mime:content-type ?TYPE. > } => { > ?R w:representation [ > mime:content-type ?TYPE; > mime:body ?BYTES ]. > } > > In those terms httpRange-14 is just: > > w:representation rdfs:domain w:InformationResource. > > (for more, see my IRW 2006 paper http://www.w3.org/2006/04/irw65/urisym > ) > > As Jonathan observes, this is frustratingly little information. > > Oh... actually, I suppose webarch does give us just a little > more information about w:InformationResource: > > "Other things, such as cars and dogs (and, if you've printed this > document on physical > sheets of paper, the artifact that you are holding in your hand), > are resources too. > They are not information resources." > > so w:InformationResource is disjoint with cars, dogs, and pieces of > paper. > > That suggests fairly strongly that it's disjoint with people, but > doesn't > say so exactly. How about Graphs? Integers? I don't find any > compelling > argument based solely on ratified specs and/or decisions of the TAG. > > -- > Dan Connolly, W3C http://www.w3.org/People/Connolly/ > gpg D3C2 887B 0F92 6005 C541 0875 0F91 96DE 6E52 C29E > > ------------------------------------------------------------ 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 Thursday, 13 May 2010 05:04:56 UTC