[whatwg] Ghosts from the past and the semantic Web
Toby A Inkster
mail at tobyinkster.co.uk
Thu Aug 28 01:51:24 PDT 2008
I consider the following to be analogous:
Presentation / Semantics / Behaviour
rel=stylesheet / rel=transformation / <script src="...">
style="..." / RDFa / on*="..."
That is, if we consider an external stylesheet linked to with
rel=stylesheet as effectively being a set of rules for mapping from
XPaths (no, CSS selectors are not XPaths, but are functionally
equivalent) to presentations, then it should be noted that
rel=transformation (i.e. GRDDL) provides us with an analogous mapping
of XPaths to semantics, and unobtrusive scripting techniques can
assign behavioural properties similarly.
Using CSS I can style any elements with class "person" as green, and
make any elements with class "name" which are descendants of class
"person" bold. With GRDDL, I can map the HTML class "person" to an
RDF class of "foaf:Person" and map the HTML class "name" to an RDF
property "foaf:name". And with some unobtrusive scripting, I might be
able to pop up an "add this person to my contacts" option when people
hover over the person's name.
But for various reasons, it is often desirable to sprinkle bits of
presentation and behaviour throughout the document. For presentation,
the style attribute is used. For behaviour, there is a whole gamut of
onthis and onthat attributes. RDFa is the equivalent for semantics.
The GRDDL and RDFa proposals basically just ask for semantics to have
the same facilities in HTML that presentation and behaviour do: that
is, an external stylesheet-like facility (GRDDL profiles and
rel=transformation) and an inline adhoc facility (RDFa's attributes
-- and there really aren't many of them compared with on*).
Bringing a good presentational layer to HTML eliminated the need to
have separate PDFs for printing. (Although many sites still do.)
Decent scripting in HTML is beginning to eliminate the need to
download and install applications -- Google and many others have been
releasing web applications which almost match desktop applications in
terms of richness of user experience. Adding a similar semantic layer
to HTML is something new -- RDF has been around for a while, but it's
always been something separate. What might it eliminate the need for
in the future? Downloads of pure data, certainly -- spreadsheets,
vCards, etc. But there are plenty of other opportunities for
semantically rich HTML -- most of which probably haven't even been
thought of yet.
Toby A Inkster
<mailto:mail at tobyinkster.co.uk>
More information about the whatwg