daniel.brickley at b...
Sat Feb 24 00:52:33 UTC 2001
So Wikis are pretty cool. DanC's recent burst of Wiki-ism had me
thinking about Wiki and the wider Web again. Just wanted to bookmark a
thought since I've a hunch an implementation would be a cheap cheap
Wiki's (structured collaborative hypertext doc collections) are islands
in the Web. For example, http://c2.com/cgi-bin/wiki?DanConnolly lets you
explore that and nearby Wikis following links between named concepts.
So I'm wondering if we could create a few RDF properties:
a property, relates a resource to a wiki name for that resource
a class representing wiki website
a relation between a wiki name and a wiki site indicating that the site
may offer Wiki info about the named resource
Thus... (no namespaces; sue me)
<wn:Person foaf:mbox="mailto:connolly at w...">
<dc:title>WikiWikiWeb front page</dc:title>
If we make the assumption (or if preferrred represent explictly) that
this wiki site allows named concepts be addressed by URL munging...
then RDFWeb implementations could figure out that a Wiki page about this
entity is available at http://c2.com/cgi-bin/wiki?DanConnolly
How does this differ from a generic 'seeAlsoSomeHTMLAt' relation? Good
question. I guess because Wiki web sites are that bit more structured,
and based around named concepts (in that sense not unlike RSS
taxonomies, RDF thesauri, topic maps or RDF schemata generally).
One avenue to explore would be conventions for Wiki's to distinguish
between classes and instances, though that'd risk over-formalising a
free flowing medium. I'd prefer to go the other route, and find ways of
taking named Wiki concepts and relating them to other RDF structures
such as shared bookmark streams.
scenario: For any given Wiki / RDFWeb combo, you
store recommendations/bookmarks/annotations associated with that named
wiki concept. Conversely, a Wiki page should be able to query an RDF
database for recent bookmarks made under that concept.
eg. I should be able to do a SquishQL query for bookmarked stuff on
http://c2.com/cgi-bin/wiki?SimulatedAnnealing (if we allow concepts to
be named by direct URIs) or on the thing whose WikiName is
'SimulatedAnnealing' or 'http://c2.com/cgi-bin/wiki?SimulatedAnnealing'.
Now I'm wavering slightly between a couple of different models. I like
the idea of Wiki names being server independent, ie. same name, two
servers. And I like the idea of using controlled names (URIs) for wiki
concepts. Not sure can have it both ways.
Anyway, that's a part-baked thought bookmarked for future use.
ps. very similar things can be said of MOOs; they're also islands in the
Web with private identifiers. http://snowball.ilrt.bris.ac.uk/~pldab/rdf-moo/
More information about the foaf-dev