[Top] [All Lists]

Re: [ontolog-forum] CL, CG, IKL and the relationship between symbols in

To: "John Black" <JohnBlack@xxxxxxxxxxx>
Cc: "[ontolog-forum] " <ontolog-forum@xxxxxxxxxxxxxxxx>
From: Pat Hayes <phayes@xxxxxxx>
Date: Sun, 6 Jan 2008 10:24:20 -0800
Message-id: <p06230917c3a6c4063edd@[]>
>I've read some of the references...I'm happy to concede that 
>'context' may be better replaced by several more specific terms, or 
>at least that we always make clear that there are many *kinds* of 
>context. And I'll keep reading...
>But I'd like to return to one part of this discussion with an 
>example 'from the field' rather than my own manufacture. I'd like to 
>get opinions about this example as it still troubles me.
>on Sat. Dec. 29, 2007 at 1:04 PM, Pat Hayes wrote:
>><huge snip>
>PH>>>The key point is, what would count as a
>>>>'context' for a context-dependent URI?
>In the following example[1], the differing 'contexts' are the 
>different web-pages upon which occurrences of a URI appear.
>PH>>> Consider
>>>>this scenario. You, sitting at your computer, use a URi to browse an
>>>>interesting website, and you send me an email telling me about it and
>>>>citing the URI. I then, sitting at my computer, two days later on the
>>>>other side of the planet, type that URI into my browser. We expect that
>>>>we will see the same website: but what do our two contexts have in
>>>>common? It might be almost nothing: the times, places, browsers,
>>>>countries, users, OSs, maybe even cultural and linguistic settings, can
>>>>be completely different. It is inherent to the Web that the contexts of
>>>>publication and of use of a URI can be arbitrarily different and far
>>>>apart on every dimension, yet the URI is supposed to retain its meaning.
>This scenario is not applicable here. We need the following 
>scenario. You, sitting at your computer, use URI-A to browse to an 
>interesting web-page upon which you see a small graphic, retrieved 
>by an occurrence of URI-CD, which refers to an assertion that the 
>web-page you are viewing is written in valid XHTML 1.0.
>This is URI-CD:  http://www.w3.org/Icons/valid-xhtml10
>What this URI is intended to denote is this assertion (from the W3C 
>help page: http://validator.w3.org/docs/help.html) "To show readers 
>that one has taken some care to create an interoperable Web page, a 
>"W3C valid" badge may be displayed (here, the "valid XHTML 1.0" 
>badge) on any page that validates."    (01)

OK, very nice example. But lets be very clear about what exactly is 
denoting what. The badge is a graphic, which is emitted by (and hence 
webarch:represents) a Web object (belonging to the W3C) which itself 
is denoted by the URI-CD (because, following the httpRange-14 
decision, no 303-redirection was done when you GET that URI). The 
badge is not itself a URI. Moreover, while it is being the object of 
the URI or URTIs which denote it, it isnt (yet) being used 
symbolically: it is simply an information object, a PNG file. But 
when used *as a badge*, ie as part of a visible Web page, then (let 
us agree) it does become a symbol, one which, as you correctly point 
out, is used indexically, since its occurrence on a Web page makes an 
assertion about that very page. But it still isn't a URI; so all that 
follows is that Web pages may contain indexical (human-readable) 
content, which is of course true. A web page containing an article 
which says "this article" contains indexical content also. Human 
languages and communication systems are rife with indexicality.    (02)

But look what happens if you try to make that assertion (about being 
a W3C valid webpage) machine-readable. Lets assume that there is a 
category - a class - of W3C-valid web pages, denoted by a 
URIreference http://www.w3.org/IconClasses/valid-xhtml10 so that what 
we need to say is that this web page is in that class, in RDF    (03)

<this web page> rdf:type http://www.w3.org/IconClasses/valid-xhtml10 .    (04)

Well, we need a URI for this web page. In your example, it was URI-A, 
right? So the RDF is    (05)

URI-A rdf:type http://www.w3.org/IconClasses/valid-xhtml10 .    (06)

and we are done. BUt now there is no indexicality, and for a very 
good reason: this assertion can be anywhere. Even if it is stored 
initially on the web page at URI-A, that is no guarantee that it will 
stay there and only there. It is intended for publication and 
transmission on the Web, and it should *retain* its meaning when that 
happens. There would be (literally) no point in having an assertion 
on a Webpage which could not be used elsewhere, as it would be 
effectively invisible.    (07)

There is another point. Even if we were to allow indexicality, that 
is a much more tractable business than arbitrary contextuality. 
Indexicality is pretty well understood, and takes only a smallish 
tweak on a standard model theory to handle it: basically, you have to 
treat the semantics as attached to the occurrence of the symbol 
rather than the symbol itself (or, equivalently, you treat the actual 
symbol, in the case of an indexical term, as being the pairing of the 
indexical with the appropriate part or aspect of the context of use: 
and indexicals indicate which part is relevant, so "here" refers to 
space and "now" to time, etc..) So, as usual, it isnt helpful to jump 
to an extremely general and poorly-understood term ('context') when a 
more precise and narrow one, on which a lot of work has already been 
done, is available.    (08)

>Now you go to URI-B and see that same graphic, because the author of 
>that web-page put an occurrence of URI-CD on his page as well. But 
>this occurrence of URI-CD is used by the author of URI-B, to refer 
>to his (author of URI-B) claim that URI-B (rather than URI-A) is an 
>"interoperable Web page".
>My claim is that the referent of each occurrence of this URI 
>http://www.w3.org/lcons/valid-xhtml10 is context-dependent upon the 
>web-page upon which it appears.    (09)

No. That URI, wherever it appears, always denotes the Web object that 
emits the graphic. The graphic itself, when used symbolically as a 
'badge', is indexical. But there are no indexical URIs anywhere in 
this story.    (010)

>In other words it is essentially indexical: it always
>refers to an assertion that "THIS page validates", where *this* 
>changes with each occurrence.
>Now the strongest objection to this I can see so far is this: The 
>URI denotes a single unique assertion    (011)

Nope. It definitely denotes the Web object which resides on a W3C 
server somewhere.    (012)

>as an uninterpreted whole, a black box, so to speak. Any further 
>interpretation of that opaque sentence, as an assertion of validity, 
>for example, is out of the scope of web-arch entirely. The 
>indexical, specifically, is only encountered later, by the client or 
>her application. Therefore, the URI denotes the same thing 
>everywhere.    (013)

Something like that, yes.    (014)

>I'm still thinking about this...
>[big snip about fingerprints]
>PH> No, I think fingerprints don't have referents at
>>all. Unless someone invents a 'language of fingerprints', but I haven't
>>seen such a thing ever proposed.
>I think this is a bit unfair since you used your rutabaga-reference 
>example. Does this imply a 'language of rutabagas'? :-) but no 
>matter.    (015)

The difference is that the rutabaga is the referent, the thing Im 
referring to, not the symbol which does the referring. In the example 
I refer to it by a gesture, rather than verbally. We subtle humans 
can refer in all kinds of ways, eg by a glance or a small nod of the 
head or even a raised eyebrow.    (016)

>PH from another post>>"I can refer to a thing without naming it. We 
>do this all the
>>>time. For example, I might pick up a vegetable in a supermarket and say to
>>>you, "Rutabaga?", using the vegetable itself as the object of my query."
>JB>>So it is with a name (in
>>>context) or a URI: The time-varying set of representations of the resource
>>>doesn't change, but *what it is* may vary.
>PH> But that is exactly what CL denies about names
>>(and the TAG about URIs). Under appropriate http-200 circumstances, a URI
>>denotes what it accesses. Which is unique, right?
>I'm claiming that http://www.w3.org/lcons/valid-xhtml10 uniquely 
>accesses a small .gif file, but that it denotes an assertion about 
>the the web-page upon which each occurrence of the URI is embedded. 
>Furthermore, I say the W3C clearly stipulates that it denotes that 
>assertion and not the .gif, in spite of REST and TAG, that it has 
>been used successfully to denote it for years, without breaking the 
>web, and that its denotation is just as usable by machine-processors 
>as by humans, since it may form the basis of a parser choice, for 
>example.    (017)

No, I think you have this wrong. The URI denotes the source of the 
image files, and that (occurrence of the) image then, when 
appropriately used, can denote the page it is residing on and say 
something about it. But it only says this when it is on that page, 
i.e. implicitly: to make this into an explicit publishable assertion 
(using RDF or OWL or indeed any other ontology language) , you need 
to use a NAME for the web page in question: and as soon as you put 
that name into the assertion, it is no longer indexical.    (018)

>Now are you saying that the CL specification, along with the TAG, 
>specifies that names must not refer to such indexical assertions?    (019)

In CL you can't refer to assertions at all, so the question is moot.    (020)

>>And under the REST model, a resource (in current jargon: information
>>resource) *is* a function from times to webarch:representations. So *what
>>it is* cannot vary, by definition, as a basic Web-architecture
>>One reaction to this is to denigrate the lack of context-sensitivity in
>>current Web communications.
>My objection is subtly different, my objection is that a science of 
>the web may show that there are already some kinds of 
>context-sensitivity in current Web communications and it is not 
>breaking anything. And this is because some kinds of context are not 
>hard to handle.    (021)

Well, you might put it better (IMO) by saying that (some) indexicals 
are not hard to handle. But the key point, why this is not a 
counterexample to what I said below, is that it is not about 
indexical *names* - URIs - but about indexicality in 'badges', which 
I will concede is a real phenomenon.    (022)

Pat    (023)

>For example, a browser knows what web-page it has got, and thus has 
>an easy time resolving the indexical.
>>But another is to simply accept that the open, global nature of the Web
>>makes it essential to presume that Web names - URIreferences - are
>>transparent (every occurrence of them have the same meaning) and to write
>>ontologies accordingly. The cost of the former - a complete re-design of
>>the entire apparatus of Web logic along lines that are yet to be
>>researched - seems to greatly exceed that of the latter, which really
>>amounts simply to following a certain discipline in inventing URI
>>references. At any rate, like it or not, this is how the SWeb is evolving.
>1. http://kashori.com/2006/07/problems-identifying-information.html
>John Black
>www.kashori.com    (024)

IHMC            (850)434 8903 or (650)494 3973   home
40 South Alcaniz St.    (850)202 4416   office
Pensacola                       (850)202 4440   fax
FL 32502                        (850)291 0667    cell
phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes    (025)

Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (026)

<Prev in Thread] Current Thread [Next in Thread>