Hi John:
IMHO (i) Martin is right regarding the (interpretation of the)
definition of rdfs:seeAlso (ii) Tim is right regarding the practical
issues thrown up by use of the wider interpretation of the range of
rdfs:seeAlso.
The question is: how to fix this for the community in general -
maybe we should have 2 relations, one with a more restricted range
than the other...?
Thanks for the nice summary and the attempt to reconcile the two views
on this. In fact, I was pretty surprised become subject to such verbal
fury ("you made that up" ;-)") for just citing the official W3C RDFS
spec.
Since there are many shop applications currently using rdfs:seeAlso to
point to product images, based on Yahoo's official GoodRelations
pattern recommendation from 2008 [1], every client MUST expect to run
into non-RDF resources when following rdfs:seeAlso. That is a simple
matter of fact.
The problem may not yet have had significant consequences since
- the product images are often rather small files and
- FOAF-related clients may rarely consume shop site data.
Yet it can lead to serious trouble in the future; hence it is good to
have this discussion now (except for the tone of the argument, to be
frank).
So fixing the code of existing clients that employ a non-standard
heuristics instead of checking the HTTP results header and evaluating
the media type and / or size of the representation of the resource
seems the only option to me.
In parallel, we can discourage people to use rdfs:seeAlso to point to
non-RDF resources in the future. It can easily be substituted by
foaf:depiction for images and foaf:page for HTML resources without
RDFa. For PDF and other human-readable resources, DC may provide
useful properties.
Best
Martin
[1] http://developer.search.yahoo.com/help/objects/product#tab2
On 13.01.2011, at 18:41, <john.nj.dav...@bt.com>
<john.nj.dav...@bt.com> wrote:
John Davies.
-----Original Message-----
From: public-lod-requ...@w3.org [mailto:public-lod-requ...@w3.org]
On Behalf Of David Booth
Sent: 13 January 2011 17:17
To: Martin Hepp
Cc: Linked Open Data
Subject: Re: Semantics of rdfs:seeAlso (Was: Is it best practices to
use a rdfs:seeAlso link to a potentially multimegabyte PDF?)
FWIW, I also agree with Martin's comments. It is the client's
responsibility to decide what to retrieve and accept:
1. The definition of rdfs:seeAlso very clearly states that "When such
representations may be retrieved, no constraints are placed on the
format of those representations."
http://www.w3.org/TR/rdf-schema/#ch_seealso
2. Only the client can know what formats and how much data it wants.
3. The HTTP protocol already provides content negotiation and HEAD
features to allow a client to find out what formats and data quantity
are available before retrieving the data.
4. There is no hard and fast distinction between RDF data and non-RDF
data. With the right de-serialization, *any* machine readable data
can
be viewed as RDF. This is not only what GRDDL does with plain XML,
but
it is inherent to RDF itself, because RDF is a data model -- not a
syntax. If the client can de-serialized from a particular format to
RDF, then the document can be viewed as RDF, regardless of whether it
can *also* be viewed as something else. (After all, n3 can *also* be
viewed as plain text.)
IMO, if there are clients that ignore available HTTP features and
blindly retrieve large quantities of data that they cannot consume,
then
those clients should be improved.
--
David Booth, Ph.D.
http://dbooth.org/
Opinions expressed herein are those of the author and do not
necessarily
reflect those of his employer.