Re: SKOS and tags (folksonomy) a bridge?

On 4/25/05, Karl Dubost <karl@w3.org> wrote:

> <skos:Concept rdf:about="http://del.icio.us/tag/photo">
> 
> <skos:prefLabel>photo</skos:prefLabel>
> <skos:altLabel>photos</skos:altLabel>
> <skos:altLabel xml:lang="fr">photographie</skos:altLabel>
> <skos:altLabel xml:lang="en">photography</skos:altLabel>
> 
> <skos:definition>Describe things related to
> photography</skos:definition>
> </skos:Concept>

Yes, I think this works, with or without something like Richard's Tag
ontology (see thread around [1]). With, you can have something like:

tag:Tag rdfs:subClassOf skos:Concept

Without, it still works nicely as seen in one of the products of
Morten's FOAF Output plugin for WordPress (thanks for the cue - I
nearly didn't mention that this week ;-)

<Concept rdf:about="#c20">
	<dc:identifier>20</dc:identifier>
	<prefLabel>Books</prefLabel>
	<rdfs:label>Books</rdfs:label>
	<inScheme rdf:resource="#scheme"/>
	<broader rdf:resource="#c23"/>
</Concept>

The label 'Books' in this case had been given by the WordPress user,
the vocab generation being automatic.

What I think is worth noting is that this approach enables the
creation and identification of multiple vocabularies, per-service
(del.icio.us, flickr) and per-user (the extract above goes in a
namespace belonging to the individual WordPress user).

It may be expeditous at times to use a common:
http://example.org/the/one/true/tagsonomy/photo

Stefano Mazzocchi's has a post around [2] discussing something like
this kind of approach (sorry, can't remember the details and the page
is currently unavailable) .

But I reckon in the long run it's likely to be more useful to keep the
concepts/tags in separate namespaces, only applying equivalence
relationships after the equivalence has been determined, rather than
starting with the assumption
http://englishblog.org/chat = http://frenchblog.org/chat

I've been playing with the same basic SKOS approach, taking
categorised channel lists (specifically Bloglines) XML and XSLT'ing it
into RDF/XML, eventually I'm hoping to insert the per-user
namespacing. The plan is to put this all in a triplestore, then
provide categorised/tagged indexes which will query using SPARQL
(details around [3]).

Cheers,
Danny.

[1] http://lists.w3.org/Archives/Public/semantic-web/2005Apr/0042.html
[2] http://www.betaversion.org/~stefano/linotype/
[3] http://dannyayers.com/archives/2005/04/19/opml-revisited/

-- 

http://dannyayers.com

Received on Monday, 25 April 2005 22:36:49 UTC