Mikael said:
> > * Simpler cross-searching of heterogeneous simple and qualified
> > sources (e.g. Edutella).
Roland said:
> Don't think so. "heterogeous" is saying: expect
> large variations of coding styles and semantic extensions of RDF(S).
>
> We're harvesting on DC stuff since 1996 and i think we know
> quite well, what "heterogenous" means in the DC world.
OK, maybe we in the DC world have learned to live with this level of
heterogeneity. But the message I dug up on www-rdf-interest (and I
imagine we could find others) suggested that working with a notion of
heterogeneity _so_ broad that it accepts that a property can take both a
plain literal as value and a resource as value is, beyond the DC world,
terminally confusing to the (potential) user [1].
Roland said:
> Additional typing is needed right away - some loop holes i
> expect with that till one really get's it clean and useful.
> One needs backwards incompatible ontologies to declare the
> range restricted DC vocabulary.
I must admit this worries me too... hence my musings a few days ago
about the possibility that really doing this properly might require a
new set of terms with new URIrefs. But I recognise that's a big job and
politically it's probably a non-starter.
Pete
[1]
http://lists.w3.org/Archives/Public/www-rdf-interest/2003Sep/0167.html
|