W3C home > Mailing lists > Public > www-tag@w3.org > December 2007

Re: The meaning of "representation"

From: Xiaoshu Wang <wangxiao@musc.edu>
Date: Wed, 05 Dec 2007 00:08:06 +0000
Message-ID: <4755EBE6.10701@musc.edu>
To: Harry Halpin <hhalpin@ibiblio.org>
CC: noah_mendelsohn@us.ibm.com, Chimezie Ogbuji <chimezie@gmail.com>, Mikael Nilsson <mikael@nilsson.name>, Tim Berners-Lee <timbl@w3.org>, www-tag@w3.org

Harry Halpin wrote:
> Xiaoshu Wang wrote
> [snip]
>   
>>  But of course, as you said, we have different definition about
>> "information".
>>     
> Well, there's about 6 or so definitions of information according to
> people working in philosophy of information, and it appears unclear if
> they layer on each other.
>   
>>>> I snip the rest (to shorten the message) because I agree your
>>>> interpretation of Shannon's theory. However, I disagree that the
>>>> assumption that the number of messages, with regard to a URI's
>>>> representations, is finite. In principle, I can use ONE bit message in
>>>> conjunction with various content types to answer all your questions
>>>> about the resource. From a communication point of view, a user do not
>>>> have a pre-established context with the resource.
>>>>     
>>>>         
>>> At a given time it would seem that the number of possible
>>> representations returned by a URI is finite. And the user does have a
>>> pre-established context with the resource, via the standards implemented
>>> by the browser, etc. etc. It may not completely determine the
>>> interpretation of the resource as regard a human user (although it might
>>> for a computer), but then....nothing does in any sort of communication
>>> when humans are involved.
>>>   
>>>       
>> Well, now you used another term "representation".  Is "representation"
>> information or a carrier of information? This is what is at debate.
>>     
> I'm using it in the webarch:representation sense (which, I might add,
> thanks to Sean Palmer for clarfiying a bit for me). 
>
> Now, one pretty good definition of information is by Bateson - it's a
> "difference that makes a difference." I'm not sure if a representation
> being information or a carrier of information makes a difference.
>
> To me, the question is - is there a difference between an information
> resource, like the novel Moby Dick, that's nature is "essentially
> information", and a web-page (webarch:representation) about Moby-Dick
> which returns representations?
>
>  Is a novel about Moby Dick an information resource? Seems so. May it
> have a web-page? Maybe. But then it might not, I could want to talk
> about "the set of all possible Moby Dick novels in the future and past",
> which would not be web-accessible. So, I guess 200 is okay there,
> although it's pretty unclear what you'd return as a representation.
>
> Like Pat, that's why I'm sort of pro-making a distinction between
> Web-accessible resources and information resources, where information
> resources are kinda unclear sometimes but Web-accessible resources mean,
> well, documents/web-pages/bytes over the Web in the form of
> webarch:representations. But defining "web-accessible" is kinda hard too.
>   
To be fair, let me try to explain how I understand Tim's information.  I 
can view tbl:Information as the result of  processing some of 
common-sense:Information.  Therefore, you don't need to request it 
because WYSWYG; and there is no need to differentiate between 
"representation" from "information resource".  In Tim's vision, all 
slash URI should be used to denote only this sort of thing.  (303 is a 
compromised version). By this vision, slash URI becomes the interface 
between human/machine and real world thing (which should be exclusively 
denoted by #URI). In other words, tim:InformationResource (I am not sure 
if it is the same as the awww:InformationResource) is one of its kind 
and it should be disjoint with any other things that we know - physical, 
temporal or conceptual.

But the implication of this model is that the information is closed.  
Hence, nothing should be further said about it .  There is no need to 
because a tim:Information is what it is.  (the reason that I made the 
Buddha analogy is out of this, not for joking but truly for 
enlightening).  Is it beautiful? Sure.  Is it practical, I don't know 
but I have serious doubt.

Nevertheless, to use the existing awww:InformationResource to explain it 
won't work (as I have argued in my article).  To say it (httpRange-14) 
helps to remove the URI's ambiguity is also a lame excuse. There are all 
kinds of semantics, and on various granularity, that need to be 
denoted.  One particular model may save you from creating one or two URI 
in one particular semantic dimension but won't do so in numerous 
others.  Besides, how can we use an ambiguous definition - 
InformationResource - to help people disambiguate?

Xiaoshu
Received on Wednesday, 5 December 2007 00:10:43 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 26 April 2012 12:47:51 GMT