[Geowanking] Re: [rdfweb-dev] locative packets
jo at abduction.org
Fri Jan 9 18:30:31 PST 2004
> I'd just subprop geo:lat and geo:long into your schema. Why force your
> document makers to always include another namespace which is super tiny
> and has terms that are within your problem domain?
Part of the spirit of the semweb, to me, is the use of others'
vocabularies and ontologies where useful and possible. The geo vocabulary
came about after several of us here and on irc started comparing our own
spatial schemas and tried to settle on a simple lowest common denominator.
Bots and aggregators are more likely to 'know about' and trust successful
RDFS or OWL models, like FOAF itself.
As much as i'd like it, i don't expect everyone running a spider or
aggregator to have an RDF query engine with enough awareness of OWL and
RDFS semantics to grok subClassOf, subPropertyOf and sameAs and
reflect that in query results. As much as i don't like it, i feel bound to
prepare for people running pure-xml- or even regex- based parsers, at
least for a while yet. And this community is perhaps open to accusations
of lack of pragmatism, logical idealism.
Also use of a stable and popular third party schema like foaf or dublin
core has better 'PR value' and doesn't involve so much buying into a
worldview or being subject to definitions and constraints on classes and
properties being suddenly changed...
> I like bounding box with low res bitmap.
chris goad's rdfgeom2d vocabulary expresses most of that, and his
pointmapper service may expand to encompass it...
cheers for the SD pointers, i hear these gaslamps are a good thing. you
probably have a point, hehe, about annotating precise locations rather
than larger areas, and given this is RDF there's nothing precluding people
from implementing support for that...
"Common sense won't tell you. We have to tell each other." -DNA
More information about the Geowanking