[
Lists Home |
Date Index |
Thread Index
]
* Simon St Laurent
|
| No, it's the RDF mindset. RDF people are perfectly happy
| manipulating URIs as identifiers, so they slap that into XML (via
| namespaces) without concern for the endless conversations and
| underspecified processing that model produces - to the labels used
| by markup, no less.
|
| RDF itself hasn't poisoned XML directly, unless Liam Quin gets his
| XMLr proposal through. RDF assumptions about the usefulness of URIs
| as identifiers is downright toxic, however, when applied to XML.
Hmmm. I think you can use URIs as identifiers without getting into
difficulties over whether you regard them as locators or identifiers.
RDF uses URIs in both ways, but the core RDF model does not make a
distinction between these two cases. Without knowledge of specific RDF
properties (essentially, application knowledge) there is no way to
know whether a specific URI is an identifier or a locator.
RDF's agnosticism about what URIs really denote is attractive in some
ways, and apparently the logicians claim that it causes no
difficulties, but it makes me somewhat uneasy for reasons I find
difficult to fully articulate.
I think I agree with your views on the symptoms (the badness of
namespaces and suchlike), but I am not sure I agree about the nature
of the disease, or even the existence of a specific disease causing
these symptoms.
Personally, I think URIs as identifiers is a great idea, and one that
has the potential to solve many crucial problems in computing. In
working with topic maps my views on what is important in information
management have started to change, and I now think much more in terms
of identity, distinctness, and sameness than I did before. I find this
extremely helpful, and URIs to be a great help in this regard.
I think topic maps provide a way for us to have this cake and also eat
it, in the sense that topic maps let you use URIs as identifiers
without losing sight of the fact that they are also locators. In RDF a
URI identifies a resource, and the nature of the resource is beyond
the scope of RDF.
In topic maps, topics represent subjects (subjects being synonymous
with resources in the RDF sense). A topic may have a URI which is the
address of the resource, in which case you can go out and download the
resource (or a copy thereof) from that URI. The URI would then appear
in the [subject address] property in the data model. Of course, that
URI then also identifies the resource, so that if some other topic has
the same URI as its [subject address] you know that the two topics
have the same resource as their subject (i.e. the topics represent the
same subject).
Of course, for subjects such a people, airports, and operas this does
not work, since these subjects do not have URIs. In this case you can
use a URI that points to a resource which describes (or indicates)
that subject (and only that subject). In this case the URI goes in the
[subject identifiers] property in the data model. If you later find
another topic that has the same URI in that property clearly these two
topics represent the same subject, since the resource only describes a
single subject.
In other words, topic maps allow you to use URIs as identifiers in a
way that does not conflict with their nature as locators of
information resources. RDF stays out of this territory altogether, and
whether that is to its detriment or not is clearly something on which
there is more than one opinion.
BTW, the published subjects work is trying to create identifying URIs
that are usable both in RDF and topic maps, but in accordance with the
topic maps view of URIs. So far it seems to be succeeding.
| See also the "patterns vs. identifiers" thread for a much more
| detailed explication of why I think this mindset has caused grievous
| and unnecessary harm to XML.
I read that, and liked the beginning, but felt that the posting
fizzled out before it really got started, which was a bit
disappointing. A full-fledged article on this would have been
interesting.
--
Lars Marius Garshol, Ontopian <URL: http://www.ontopia.net >
ISO SC34/WG3, OASIS GeoLang TC <URL: http://www.garshol.priv.no >
|
