[microformats-discuss] Microformats and the Semantic Web

[Danny Ayers]

On 8/8/05, Tantek Çelik <tantek at cs.stanford.edu> wrote:

One or two comments (in general I agree). 

> Needless to say, I take the former position, with the additional statement
> that microformats are easier to both define, deploy, and *present* on
> today's Web, thus, given equivalent semantics, it makes more sense to
> publish microformatted data rather than RDF/XML on the Web.

I would suggest that the appropriate format depends on a few factors -
primarily the nature of the data. XHTML's basically a doc format for
human-readable knowledge representation (after rendering), whereas
RDF/XML is designed for RDF graphs/machine-readable knowledge
representation (with or without rendering). Some of the more complex
schemas/ontologies (e.g. the life sciences stuff) would be hellish to
express as microformats, and in many cases the data wouldn't be of any
real benefit presented directly to the user as XHTML (e.g. map data).
Yet there is still huge potential benefit to publishing the stuff on
the Web.

There are other practical issues, for example the other day I wanted
to get the whole of my blog data into a triplestore, so I serialised
it as RSS 1.0 (RDF/XML), with the extras provided by FOAF Output and
read that directly (ok, after a little cleaning) into the triplestore.
At some point I'll expose the whole lot on the Web so anyone else can
reuse it as they wish. An alternative might be to put the MySQL dumps
online, but that would make reuse more difficult for anyone without
MySQL. All this data (more or less) is already available online as
HTML, but reuse would involve crawling and interpretation to somehow
get the data back into a machine-processable form. And this stuff's in
a relatively flat structure, if there was interlinking it'd be harder
still.

> What data model and data structures you use *inside* any particular
> application should be determined by whatever is most efficient for you the
> developer.  If for you, that's RDF that's fine, if that's C structs that's
> fine, If that's a relational database, that's fine too.  No need to argue
> about which *internal* representation is better, just use the one you think
> will best help you get your job done.

There I agree with the immediate point, but probably see the
implications a little differently. One of the motivations behind the
Semantic Web is to enable data sharing at a level above syntax. In
other words using an *external* representation that applications can
share. This is a fairly natural progression from the current Web,
where URIs provide globally-shared names for resources. The specific
serialization format is secondary -  whether it's RDF/XML or a
microformat.

> This focus on existing data, and thus the lowering / minimizing of barriers
> to re-using it, IMHO is one of the key distinguishing principles of
> microformats over the Semantic Web, both in theory (the microformat
> principles), and in practice (the process for defining microformats, based
> on existing *visible* data on *today's Web* and existing schemas both
> implied and explicit).

Hmm, ok, but I'd add that there's a lot of data tucked away in
databases that could be exposed more easily than through templating to
XHTML.

> > Where there may be a
> > difference is that the Semantic Web in the sense of the W3C initiative
> > is backed by a logical formalism and Web-global data model (RDF).
> 
> Which some folks see as a foundation, and others as a barrier.

The latter folks are missing a couple of significant points. It
doesn't necessarily cost anything to take advantage of the formalism.
I haven't a clue about the operational semantics of Python, but I can
still use "print 2+2". A Web-global data model means that someone can
interop/reuse material without any additional agreement -
"http://xmlns.com/foaf/0.1/name" will always mean the same thing,
"name" might not.
 
> Microformats essentially ask:
> 
> Can we do more (practical use and applications)
> with less (logical formalism, formats, namespaces, etc.)?

Sounds reasonable ;-)

Cheers,
Danny.


-- 

http://dannyayers.com