Re: gimmee some data!

On Tue, 2009-06-16 at 07:30 -0400, Ross Singer wrote:
> On Tue, Jun 16, 2009 at 4:24 AM, Toby Inkster<tai@g5n.co.uk> wrote:
> Definitely.  Since it's not an absolute that the coverage of ISBNs are
> 1:1 (in either direction, but definitely from yours to here) -- how to
> determine what actually exists?

In my opinion, that's a non-issue. A 404 does not mean that the book
does not exist, just that the web server has no information about it. So
the meaning of, say:

 http://dilettantes.code4lib.org/LODThing/isbns/0062515862

is well-defined, even if no data is returned, and I can safely say that
http://purl.org/NET/book/isbn/0062515862#book is owl:sameAs it. (Though
in fact, there should never be a book with that ISBN as the checksum
doesn't work out.)

However it would help if you used different URIs for the bibo:Book
instances and the RDFa documents which describe them. q.v. httpRange-14.
That way, you could distinguish between the pages and the book they
describe - e.g. to say that the book exists but the page does not. It
would also allow some basic statements like

<http://dilettantes.code4lib.org/LODThing/isbns/0062515861>
  foaf:maker
    [ foaf:name "Ross Singer" ] .
<http://dilettantes.code4lib.org/LODThing/isbns/0062515861#book>
  foaf:maker
    [ foaf:name "Tim Berners-Lee" ] .

> It seems to me if we're talking about the same ISBN, owl:sameAs is
> accurate, but I don't have strong enough opinions to back that up in
> an argument.

We are both using FRBR - you directly, and me via a subclass - and
thought we might be doing our modelling at different levels in FRBR's
hierarchy. But it seems we're both treating an ISBN as representing an
frbr:Manifestation, so it should be safe to do.

I'll add the owl:sameAs stuff on my side at lunch time.

-Toby

Received on Tuesday, 16 June 2009 11:55:32 UTC