[uf-discuss] Re: GRDDL Profile for Links

[Danny Ayers]

Hi Jonathan,

On 30/10/2007, "~:'' ありがとうございました。" <j.chetwynd at btinternet.com> wrote:
> Danny,
>
> I enjoyed the blog and the stripped rdf and html even more. tx.
>
> but what's your intention, don't many if not all web2.0 strip links
> or other info out of web pages?

Not sure I follow... the source material is regular HTML, though
typically material aggregated over RSS/Atom does lose a lot of
potentially reusable data (although basic GRDDL facilities for these
formats are available).

The intention is to make the data in the HTML pages available to other
tools at minimal cost. A decent number of RDF tools now support GRDDL,
and for the publisher the RDFification is (less than) a one-liner.

Kingsley pointed me to OpenLink's RDF browser viewing the data -
http://tinyurl.com/28e5mm - has SPARQL query facilities built in, etc
etc.

> I'm particularly interested in developing user tools for repurposing
> content including but also beyond CSS. for example http://
> www.peepo.co.uk/mybbc/hints.html demonstrates how BBC news text and
> links without images are not displayed using CSS.

Hmm, interesting. I'm not sure about the notion mentioned there as
"less links" per se, it sounds like the preference is more like "fewer
confusing visible links" - more of a styling thing (hence the CSS..?).

Funnily enough, the bug correction Reto for my representation of
@titles, making the link a resource in itself (as well as the linked
resource) could be very handy for prioritising links for visibility.
Maybe some kind of annotation facility to rate the links on a page,
then the original page could be passed through a filter which would
assign visibility according to those ratings.

I'm not really up to speed on accessibility, but would imagine GRDDL
would have a lot of potential. The RDF works closer to bare concepts,
without so much prose to muddy the cognitive picture. A rendition of
the information could go 'below' the text, though I can't really
imagine what concrete form such semantic stylesheets might take. Tools
like Fresnel suggest a way to implement.

Whatever the stuff looked like consumer-side, the fact it doesn't make
any major demands on the publisher (that sub-one-liner) should help
avoid the kind of adoption issues that would otherwise arise.

Cheers,
Danny.

-- 

http://dannyayers.com