Lists Home |
Date Index |
John Cowan wrote:
>>(a) have a built-in way to disambiguate between what the resource "is"
>>and what it's "about"
>>(b) use different names.
>>I prefer (b), and would encourage people to use something like
>>http://www.heritage.org/Shakespeare for the person and the URI above for
> The trick with (b) is that you are then impaled on this dilemma:
> (b1) Dereferencing "http://www.heritage.org/Shakespeare" returns something,
> (b2) Dereferencing "http://www.heritage.org/Shakespeare" returns nothing.
> With (b1) you now have yet another document and have to bifurcate again;
> with (b2) you get a lot of surprised people who dereference an http: URL
> and get nothing, just like with namespaces. Given this choice, I like
> (a) better.
I must be missing a circuit in my brain - lots of smart people worry a
whole lot about this cluster of problems and they've just never affected
me - maybe a function of the kind of application I write. For example,
neither (b1) nor (b2) give me heartburn. If I have a resource that I
claim represents Shakespeare (probably for the purpose of making
statements about him) then if dereferencing returns something, it better
plausibly be a representation of the late bard. For example, it could
be the content of the .jpg file. If I claim that the URI
.../Shakespeare.jpg identifies a picture of some particular person, then
any representation should be of the picture.
This seems to be the same thing that Simon is talking about - I could
write an RDF assertion along the lines of
http://.../Shakespeare.jpg IsARepresentationOf http://.../Shakespeare
which seems perfectly sensible.
Anyhow, what would you propose for (a)? -Tim