W3C home > Mailing lists > Public > public-lod@w3.org > October 2011

Re: [HTTP-range-14] Hyperthing: Semantic Web URI Validator (303, 301, 302, 307 and hash URIs)

From: Jonathan Rees <jar@creativecommons.org>
Date: Mon, 17 Oct 2011 09:25:35 -0400
Message-ID: <CACHXnapah-8qzt_omVim4ELP1jMBpAy_qmbF0C0cv76XM50odQ@mail.gmail.com>
To: Yang Squared <yang.square@gmail.com>
Cc: public-lod@w3.org
I'm not sure what your script is supposed to do. If you give it

http://dx.doi.org/10.1016/j.jbi.2008.03.004

it says that "identifies" a Real World Object or Thing. That seems ok,
since documents are arguably real, but this is completely
uninformative.

If you give it

http://xmlns.com/foaf/0.1/Person

it also says the referent is a Real World Object or Thing. To say that
classes like this are "real world objects" is tenuous at best. They
are not localized in space, you cannot touch or measure them, etc. And
their claim on Thingness is also tenuous. OWL DL, for example,
explicitly separates Things and Classes - i.e. Classes (like
foaf:Person) are never Things in DL. But this may be a matter of
definition (perhaps you should give citations so people know how you
are using troublesome words like "Thing").

If you give it

http://www.w3.org/TR/1999/REC-html401-19991224/

it gives a different message - it says the URI identifies an
Information Resource. But the DOI did too - as far as what the
referent is, the two URIs are extremely similar and should classify
similarly. So if the tool is meant to classify what's referred to, the
message ought to be identical to the DOI case.

My guess is that the script is trying instead to explain the *way* in
which you think the "identification" is to be determined - which
practice or rule applies. According to the httpRange-14(a) rule the
2xx URIs "identify" information resources (a generalization of
"document" that includes images and so on, and accounts for variation
between particular retrievals) - usually this is taken to mean that a
retrieval-enabled URI "identifies" the particular information resource
found at that URI, not just any old URI. When a hashless URI is used
in some other way usually a non-retrieval-enabled URI (such as a 303
URI) is used. I'm not aware of any normative text on how 302 or 307
relate to "identification", so you just need to go on what practice
seems to be.

Remember that 303 says nothing whatsoever about what the type of the
referent (and 200 is usually understood as saying something quite a
bit stronger and more interesting than that it *is* an "information
resource"). The description you get by 303-redirecting says what the
referent is supposed to be, and it can say anything at all - it can
even say that the URI refers to an "information resource" retrieved at
a *different* URI (as in the DOI case).

Tools like this are more useful when they provide not just some
judgment but also the justification, in terms of what was found and
what is specified, for any particular judgment. (The W3C HTML
validator does this really well.) This is especially important when
there are that standards status of much of this practice is so
peculiar. If you're going to be transparent and accountable you will
need to expose a more detailed and nuanced story.

Jonathan

On Mon, Oct 17, 2011 at 6:41 AM, Yang Squared <yang.square@gmail.com> wrote:
> Following the HTTP-range-14 discussion, we developed a Semantic Web URIs
> Validator named Hyperthing which helps to publish the Linked Data. We
> particularly investigated what happens when we temporary and
> permnent redirect (e.g. 301 and 302 redirections) of a Semantic Web URI (303
> and hash URI).
> http://www.hyperthing.org/
> Hyperthing mainly functions for three purposes:
> 1) It determines if the requested URI identifies a Real World Object or a
> Web document;
> 2) It checks whether the URIs publishing method follows the W3C hash URIs
> and 303 URI practice;
> 3) It can be used to check the validity of the chains of the redirection
> between the Real World Object URIs and Document URIs to prevent the data
> publisher mistakenly redirecting between these two kinds. (e.g. it checks
> against redirection which include 301, 302 and 307)
> For more information please read
>  Dereferencing Cool URI for the Semantic Web: What is 200 OK on the Semantic
> Web?
> http://dl.dropbox.com/u/4138729/paper/dereference_iswc2011.pdf
> Any suggestion is welcome.
Received on Monday, 17 October 2011 13:26:04 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 16:21:17 UTC