Re: Last minute input to discussion re 'on the boundaries of content negotiation in the context of the Web of Data'

Michael Hausenblas wrote:
> Dear TAG members, Tim, Richard,
>
> Short version: an attempt to defined non-information resources without using
> non-information resource ;)
>   
I am not sure your proposal will work (see below).  But here is the 
fundamental question: what is the purpose?  

Unlike if we know "x is a Human", we can infer that "x has (at least) 
one head", I wonder what more "x is an IR" tells me except "x is an IR". 

The original intension of defining IR, I believe, is to avoid ambiguous 
identity.  But the solution is always using different URIs for different 
things. Defining something like IR does not help.
> Long version:
>
> In order to contribute in a constructive way to the discussion around
> information resources, non-information resources, CN and 200/303 issues, I'd
> like to propose the following, which may be taken as an input to my
> previously raised issue around the boundaries of CN in the Web of Data use
> case. To additionally address emerging usage of RDFa, this area has been
> taken into account as well.
>
> Please note that this is *not* a philosophical approach and I try to ground
> all my terms and only use *existing* protocols, definitions, etc.:
>
> ===
> Axiom 1) A URI containing a fragment identifier as of RFC3986 [1] identifies
> a 'thing', that is, a resource which essential characteristics are not
> conveyed in a message as opposed to an information resource as of AWWW1 [2],
> unless otherwise stated.
>   
This gives a syntactic definition of "Thing" (which I believe is non-IR 
thing).  It is no different to name it as, say "hash-thing".  Hence, we 
can infer from a "hash-thing" that "not all essential characteristics is 
conveyed in the message" so it grants "hash-thing" an open world 
semantics.  But the implication is that "slash-thing", i.e., things 
denoted by a slash URI, should not be talked about because it is 
closed.  Otherwise, what is the difference between "hash-thing" (IR) or 
"slash-thing" (non-IR).

Furthermore, the Web is defined by three things -- URI, Representation, 
and Resource(things).  In other words, these are the three things we 
know about the Web at its fundamental level.  The syntactic definition 
of a URI can be used to distinguish these three entities (it would be 
ideal and convenient if we do so, though not necessarily). Using a 
syntactic definition for anything else would suggest that either (1) the 
Web requires additional founding concept or (2) it would violate the 
URI's opacity principle.
> Axiom 2) Iff the media type of a representation obtained by dereferencing a
> URI (that is, performing an HTTP GET on the URI) defines the semantics of
> the fragment identifier, the resource is an information resource as of [2];
> this is the case 3 in [3].
>   
Define the "define"?
> Axiom 3) Iff the authoritative party as defined in section '2.2.2.1. URI
> ownership' of [4], that is the one who can claim URI ownership, explicitly
> states that fragment identifier semantics throughout different
> representations are sufficiently consistent, the resource is an information
> resource as of [2]; this is the case 1 in [3].
>   
Another ambiguous word "sufficiently"...
> Axiom 4) An authoritative party can explicitly state fragment identifier
> semantics consistency by using POWDER's describedby property as of [5] along
> with HTTP Link: header as of [6] or by embedding RDF as of [7].
>   
This is even more trouble-some to me.  If the IR makes as a foundation 
concept of the Web, this axioms essentially making POWDER, HTTP LINK, 
RDFa at the same fundamental level of the AWWW. 
> ===
>
> Please note the following: The intention is to keep the current definition
> of IR as of the AWWW1 and clearly define what else is possible, hence to
> make related (sometimes questioned and IMHO underspecified definitions such
> as for example found in httpRange-14) more usable in a practical context.
> The axiom 1 actually asks people to use frag-ID-URIs as the (one and only)
> default to identify 'things', however, axiom 2 - 4 allow to create
> exceptions based on an explicit set of actions. Further, axiom  allows
> (currently) two ways to 'announce' exceptions, on the HTTP layer or on the
> representation layer - this is open to discussion and should/will be
> extended.
>
> @Tim: Do you think this helps in getting closer to a written explanation of
> your often articulated thoughts re 'sameness' of information obtained from
> resources, as e.g. in [8]? And also: does this solve our issue wit RDFa as
> discussed in [9]?
>
> @Richard: Are the axioms consistent with the outcome of your analysis in
> [10]?
>
> Cheers,
>       Michael
>
> [1] http://tools.ietf.org/html/rfc3986#section-3.5
> [2] http://www.w3.org/TR/webarch/#def-information-resource
> [3] http://www.w3.org/TR/webarch/#frag-coneg
> [4] http://www.w3.org/TR/webarch/#URI-collision
> [5] http://www.w3.org/2007/05/powder-s#describedby
> [6] http://tools.ietf.org/html/draft-nottingham-http-link-header-03
> [7] http://www.w3.org/2001/tag/doc/selfDescribingDocuments#UsingRDFa
> [8] http://chatlogs.planetrdf.com/swig/2009-02-09#T15-09-20
> [9] http://esw.w3.org/topic/RDFa_vs_RDFXML
> [10] http://lists.w3.org/Archives/Public/semantic-web/2007Dec/0157.html
>
>   

Received on Friday, 20 February 2009 15:03:21 UTC