[uf-discuss] Hello / Semantically-Interlinked Online
Tantek Ç elik
tantek at cs.stanford.edu
Mon Nov 7 08:51:46 PST 2005
Thanks for providing some very good context and background. I wanted to
touch on just a few of your points.
On 11/7/05 8:22 AM, "Danny Ayers" <danny.ayers at gmail.com> wrote:
> I honestly don't know the best approach to take in terms of
When in doubt, first read the process:
And if there are still doubts, ask a specific question about whatever part
of the process seems unclear.
> David's hAtom does compromise between Atom's way of
> doing things and XHTML's
In terms of David's work on hAtom, I've yet to see any real "compromise"
from a "schema" level, and that's what counts here.
> so reuse of the terms he's using may well be
> appropriate for SIOC. SIOC could potentially be expresssed in XHTML
> pretty much however you like, with GRDDL as the mechanism for
> extracting the data.
The more general (and no intention to standardize) approach of using
semantic class names works for this.
> This approach is generally consistent with the
> mechanics of microformats, if not necessarily the initiative's
> process & practices.
Microformats processes, principles, and practices are what make microformats
Without that you just have proper use of semantic XHTML, class names etc.
All of which are a good thing, and necessary, but certainly not sufficient
to design a standard interoperable microformat which is easy to use etc.
> Obviously gelling with these would be desirable.
If a problem is already solved by a microformat, then there is actually a
strong desire/motivation to NOT invent another microformat to do so.
> On the issue of namespaces & prefixes - the general approach taken in
> the context of microformats (correct me if I'm wrong guys) is not to
> attempt to generalise to the extent of IanD's RDF-in-HTML, rather to
> define more domain-specific formats.
Or put even more bluntly.
In general, microformats rejects the use of explicit namespace prefixes in
documents as unnecessary for solving the 80/20 of problems that microformats
seeks to solve. Note: I'm not saying that to open a discussion, and in
fact, we've had to terminate such ratholes in the past.
> Disambiguation of names
> (attribute values) within a document is achieved through (for want of
> a better phrase) Intelligent Design, i.e. the 'approved' microformats
> are constructed with the prior intention that naming clashes won't
A better way of putting this is that the disambiguation is achieved through
the social design of having a *very* open community (microformats.org) which
is imminently findable by anyone else doing work in the area on the Web
(this would not have been possible 10 years ago), and whose principles
explicitly forbid reinvention or duplication of formats and explicitly
encourage reuse of microformats as building blocks.
> Disambiguation on the Web at large is achieved through the use
> of profile URI(s), (<head profile="htttp://...">), effectively
> namespacing the vocabulary in a lightweight, authoring-friendly
Yes, Danny's right. Though we depend on the social construct of
microformats to avoid such problems in practice, microformats also use XMDP
profiles to explicitly define vocabularies for formats. And XMDP's URLs for
specifying terms is compatible with those used by RDF, with "#term" at the
end. This is by design, and was one of the earliest examples of
microformats reusing a convention from another format as it were.
More information about the microformats-discuss