Re: Resource identity of GETable URIs [Was Re: the mistake I made! ]

The 'identity' idea is a red herring - which is good because I really
don't know what it means.

The only practical reason to pursue httpRange-14 is the desire to
write metadata, and the only logic or philosophy that we need is
whatever is absolutely required to enable the production and
consumption of meaningful metadata.

To this end I agree with David that toucan-document chimeras are not
necessarily so bad. But they would be awful if toucan axioms forced
two such chimeras to be identified that shouldn't be. For example,
suppose toucans had unique names "Fred" "George" and so on (i.e.
toucan-name is a functional property), and toucan-chimera 1 had
toucan-name "Fred" and toucan-chimera 2 also had toucan-name "Fred".
If that caused me to conflate chimeras with distinct document-natures
that would be very bad - my graph would contain incorrect information.

Making IR and toucan disjoint might help, but a disjointness axiom is
neither sufficient (as in the CC REL case) nor necessary.

I don't think the bad situation is likely since (according to my
conjecture based mostly on reading between the lines) the people who
create toucan-document chimeras are the same as the ones who wouldn't
assert that their properties are functional.

If I care about the documents and am lucky, I'll be in ignorance of
toucan axiomatics and I'll be blissfully unaware of the equation - I
probably won't even trust the functional property axiom source in the
first place. But it would be nice not to leave this to chance. One
possible solution: Tell people to feel free to create toucan-chimeras,
but if you do, please don't use logic at the same time, just do linked
data.

Jonathan


On Fri, Mar 4, 2011 at 3:13 PM, David Booth <david@dbooth.org> wrote:
> On Tue, 2011-03-01 at 13:41 +0000, Nathan wrote:
> [ . . . ]
>> It's vital to anybody that wants to use layer 3, to have this
>> distinction Tim promotes, the guys on layer 2 stuff, like Roy and Hixie
>> don't hit that need, the guys on layer 3 like us and Tim, do hit it.
>
> I disagree.  I think Tim's view *attempts* to ensure that things are
> named unambiguously, or perhaps "encourages people to *try* to name
> things unambiguously", but as Pat Hayes has pointed out on several
> occasions,
> http://www.ibiblio.org/hhalpin/homepage/publications/indefenseofambiguity.html
> I think it is inherently *impossible* to ensure that things are named
> unambiguously for all applications.  It is always possible to make ever
> finer distinctions, so a URI whose resource identity is unambiguous to
> one application may well be ambiguous to another application that
> requires finer distinctions.
>
> Bottom line: While many applications may require that "GETtable URIs
> refer to information resources and not cars", others do not, and the
> success of the semantic web does *not* depend on making that particular
> distinction.  Nonetheless, I *do* think it is useful to be able to talk
> about which URIs make this distinction and which do not, i.e., which
> resources can be considered disjoint from the class of IRs.
>
> [ . . . ]
>> There are three layers of the web,
>> 1: a web of machines
>> 2: a web processes which we transfer data to/from and ask to do things
>> 3: a web of things talked about or referenced in information, which
>> includes things on layer 1, layer 2, and everything else one can conceive.
>
>
> --
> David Booth, Ph.D.
> http://dbooth.org/
>
> Opinions expressed herein are those of the author and do not necessarily
> reflect those of his employer.

Received on Friday, 4 March 2011 20:46:17 UTC