W3C home > Mailing lists > Public > www-tag@w3.org > September 2004

RE: new text for Information Resource (section 3.1)

From: <Patrick.Stickler@nokia.com>
Date: Thu, 9 Sep 2004 19:49:16 +0300
Message-ID: <1E4A0AC134884349A21955574A90A7A50A1CDD@trebe051.ntc.nokia.com>
To: <sandro@w3.org>
Cc: <www-tag@w3.org>



> -----Original Message-----
> From: sandro@roke.hawke.org [mailto:sandro@roke.hawke.org]On Behalf Of
> ext Sandro Hawke
> Sent: 09 September, 2004 17:37
> To: Stickler Patrick (Nokia-TP-MSW/Tampere)
> Cc: www-tag@w3.org
> Subject: Re: new text for Information Resource (section 3.1) 
> 
> 
> 
> > I suspect that your definition of "state" is not quite the 
> same as mine.
> 
> Looking over some definitions for "state" [1], I disagree.  We're
> using the word "state" in essentially the same way, I'm just thinking
> that the *entire* state needs to be transmitted, and you're thinking
> it's fine to transmit some particular information which is in some
> way extracted or derived from the complete state.
> 
> AWWW says:
> 
>      Note: In this document, the noun "representation" means "octets
>      that encode resource state information". These octets do not
>      necessarily describe the resource, or portray a likeness of the
>      resource, or represent the resource in other senses of the word
>      "represent".
> 
> not saying either "...encode all resource state information..."  or
> "...encode some resource state information...".   If I were a judge
> trying to guess the intent, I'd probably read "some" into there, but
> when I'm thinking "all" I can read that text without noticing any
> contradiction.   Superb weasel text.
> 
> > If I choose to denote the actual coffee-maker with a URI, 
> and provide
> > a representation that reflects the "state" of the coffee-maker as
> > "on", then I think that's just fine.
> 
> So I use firefox to visit http://tag-coffee-maker.example.com and I
> see the text, in black on a white background, saying "On".  And now I
> have some evidence that the TAG coffee-maker is on.  
> 
> Obviously http://tag-coffee-maker.example.com identifies a web page
> which tells people whether the coffee maker is on or off.  To say the
> URI identifies the coffee-maker itself is sophistry.
> 
> If we asked people to do a thorough investigatation of
> http://tag-coffee-maker.example.com and tell us who created it and to
> describe its coloration, would they tell us about the creation and
> coloration of the coffee-maker or the web page?  I'm rather sure they
> would describe the web page, unless there was some very confusing
> labeling on or near the page [2].   Shall we do the experiment?
> Maybe we could get each willing reader to try it with four random (not
> TAG-savvy) co-workers.....
> 
> 
> 
> So let me propose the text which says more clearly what I think
> you mean:
> 
> (1 -- minimal change to what I think Patrick means)
> 
>      Note: In this document, the noun "representation" means "octets
>      used to convey information".  These octets are typically
>      assocated in some useful way with a resource they are said to
>      "represent", but they do not necessarily describe the resource,
>      or portray a likeness of the resource, or represent the resource
>      in any particular or standard way.
> 
> or rephrased
> 
> (2 -- rephrased to be bone-crushingly clear) 
> 
>      Note: In this document, the terms "representation" and
>      "represents" are used in a specialized way to talk about the
>      relationship between a resource and a sequence of octets which
>      are said to represent it or to be a represention of it.  The
>      relationship between a resource and its representation is in
>      general not constrained, and other meanings of the word
>      "representation" should not be taken to indicate what the
>      relationship might be.  The representation does not necessarily
>      describe, depict, symbolize, or otherwise relate to the resource
>      in a way which one could know without knowledge of the
>      specifications governing the context in which the representation
>      occurs.
> 
> In contrast, I would prefer something like this:
> 
> (3 -- what I would like)
> 
>      Note: In this document, the terms "representation" and
>      "represents" are used in a specialized way to talk about the
>      relationship between a resource and a sequence of octets which
>      are said to represent it or to be a represention of it.  The
>      representation of a resource is simply its encoding in a standard
>      format.  In some cases the encoding may be a straightforward
>      serialization or marshalling of structured data, but lossy
>      encoding formats (like JPEG) can also be used to create a
>      representation of a resource.  Only resources which are
>      purely information can be encoded in bits, so only these
>      resources (called "Information Resources") can have
>      representations.
> 
> Cheers.
> 
>      -- sandro

I find myself generally comfortable with either 1 or 2 above. I would
not be happy with 3.

I think this all boils down to how (or whether) the web and semantic
web will have a seamless integration. The semantic web allows one to
talk about anything, regardless of its accessability on the web. Yet
I consider it the greatest win that the web and semantic web would have
a shared set of identifiers in that anything described on the semantic
web could also be provided representations on the web -- including
coffee makers and dogs.

Definition 3 above takes the (IMO overly restrictive) "document centric"
view that is not sufficiently generic to allow applying both the
web and semantic web layers to their full potential given a shared
set of resource identifiers.

Patrick


Patrick


> 
> [1] http://dictionary.reference.com/search?q=state
> [2] 
http://www.georgetown.edu/faculty/irvinem/visualarts/magritte-pipe-sm.jpg
Received on Thursday, 9 September 2004 16:49:31 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 26 April 2012 12:47:28 GMT