Raw

A piece in IEEE Intelligence Systems in which Nigel Shadbolt, Wendy Hall, and Tim Berners-Lee give an update to the 2001 Scientific American article. Avaliable as pdf. Harry Chen has a good synopsis.

I might be getting oversensitive, but the references to folksonomies niggled me a bit. Seems to emphasise the differences (which is the path to the false dichotomy of upper- versus lower-case semantic web) rather than highlighting the overlaps. Here's what I mean.

cat is a string

http://del.icio.us/tag/cat identifies a resource, the set of things tagged "cat" at del.icio.us - i.e. the tag for the concept associated with the string "cat" at del.icio.us, which suggests -

<http://dannyayers.com/tags/cat> rdf:type skos:Concept;

                                 skos:prefLabel "cat" .

or using the Tag Ontology to be more specific -

<http://dannyayers.com/tags/cat> rdf:type tag:Tag .

That URI identifies the concept associated with the string "cat" at dannyayers.com  <http://dannyayers.com/tags/cat> skos:broader <http://del.icio.us/tag/cat>.

It's also a specialization of my concept of cat: 

<http://dannyayers.com/tags/cat> skos:broader <http://dannyayers.com/concepts/cat>.

My cat tag is more specific than del.icio.us's.

< http://semtext.org/pets/profile.xml>  tag:taggedWithTag <http://dannyayers.com/tags/cat> .

That's Sambuca's profile, I'll tag it with my tag "cat"

Thanks to someone (?) at Mindswap, Sparql has a profile now too, I guess that could be tagged the same way too.

Both the profile I did for Sambuca and Sparql's use pet ontologies. These are talking directly about real-world things, whereas the tagging stuff is more about the concepts of things. Looks like their ontology isn't far from to mine, my Cat is an owl:equivalentClass to their Pet Cat.

I've not been following closely, but last time I looked there was a little uncertainty over how best to make direct associations between SKOS concept hierarchies and RDF/OWL classes. I think I'd opt for something like:

pet:Cat x:concept <http://dannyayers.com/concepts/cat>. 

After all that, cat is still but a string, and there is a lot of informal, sociological and statistical aspects to folksonomies. But tagging is still an artifact of knowledge representation that can be used in concert with more formal approaches like OWL ontologies. Tagging adds value to the SemWeb, the SemWeb adds value to tagging. Quo vadis.