W3C home > Mailing lists > Public > semantic-web@w3.org > March 2005

Re: Tag ontology RFC

From: Danny Ayers <danny.ayers@gmail.com>
Date: Fri, 25 Mar 2005 01:17:32 +0100
Message-ID: <1f2ed5cd050324161778237345@mail.gmail.com>
To: Richard Newman <r.newman@reading.ac.uk>
Cc: Seth Russell <russell.seth@gmail.com>, semantic-web@w3.org

Just a 1am thought - 

When I tag a resource http://nosh.org with del.icio.us "food" I'm
saying that the resource is a member of a class of things tagged with
"food" (at del.icio.us).

We have classes! The string "food" still needs to be associated with
the class, but that's not a big deal:

del:food rdf:type rdfs:Class

http://nosh.org rdf:type del:food
del:food  rdfs:label "food"

I can't (at this point in time) see a problem with generating a class
for every tag in del.icio.us's vocabulary, unless they allow odd
characters.

Although it might give the DLer's nightmares, an alternative could be:

http://nosh.org rdf:type _:f
_:f  rdfs:label "food"
_:f x:scheme http://del.icio.us

Cheers,
Danny.
 


On Thu, 24 Mar 2005 23:24:07 +0000, Richard Newman
<r.newman@reading.ac.uk> wrote:
> 
> Seth,
>    That's a very interesting visualisation; thank you.
>    I'm not sure that an RSS1.0 Item is the same as a Tagging; the
> intention of a Tagging is solely to capture the n-ary relationship
> between a tag, an agent, a resource, and a date (and anything else),
> which together make up a particular tagging of the resource by the
> agent. They're certainly similar in some aspects (being dated
> annotations of a resource), but I'm not sure if the semantics are
> sufficiently similar to warrant the same vocabulary.
> 
>    I'm not sure about taggedWithTag at all; I kept it in because it's
> simple, but it (being only a triple) doesn't allow us to capture
> who/when information, so you have to rely on the tag URI to capture
> some of that. Ideally we'd only use the reification, but both
> directions make some amount of sense for this simple triple.
> 
>    You might well be right about tag being the wrong way round; I
> modelled it from a usage point of view, where typically outbound links
> are more informative (i.e. "Richard is tagged with 'SemWeb'", allowing
> you to discover from a resource the tags to which it is linked), but
> conceptually the inverse is more sensible ("this Tagging links 'SemWeb'
> to Richard", finding resources from tags). In the real world you'd have
> to go both ways, so it's still up in the air.
> 
>    I'll have a further look at this in the morning (when I'm sober).
> 
>    Thanks,
> -R
> 
> On Mar 24, 2005, at 21:23, Seth Russell wrote:
> 
> >
> > Richard,
> >
> > It's hard for me to grok an ontology by just looking at text, so i
> > mentographed it; see  http://robustai.net/folksonomy/Tag-ontology.html
> > .  Then too its nice to see an example of it use used in a real
> > Tagging; see
> > http://robustai.net/folksonomy/aTagging.html
> >
> > You might want to compare the example with the RSS feed that you get
> > back from delicious,  (see the RSS button at the bottom of the page
> > for that tag).   After working with this i am pretty convinced that a
> > Tagging is a RSS item.
> >
> > I think that "taggedWithTag" and "tag"  are arrows drawn in the wrong
> > direction.  I suggest staying close (if not identical) to the RSS.  I
> > drew the arrows that i would prefer in red.
> >
> > If you like my choce of "tags.ontology", you might tag you items with
> > it and we can use del.icio.us to focus this discussion.
> >
> > Thanks for the dialogue
> > Seth Russell
> >
> >
> >
> >
> >
> > On Wed, 23 Mar 2005 13:27:07 +0000, Richard Newman
> > <r.newman@reading.ac.uk> wrote:
> >> Danny, Seth, sw@w3:
> >>    I've thrown together an early draft of a tag ontology. There's
> >> nothing synset-ey in there yet (other than using SKOS ConceptSchemes),
> >> but it all ties gently in with the SKOS vocab. I'd be looking to
> >> extend
> >> this integration if possible.
> >>
> >>    Danny, your email re the WordPress plugin was very interesting; I
> >> think it follows through with this, in that you could do the
> >> following:
> >>
> >> <item rdf:about=" ... ">
> >>    <dc:subject>Programming</dc:subject>
> >>    <tags:taggedWithTag rdf:resource="
> >> ...my/personal/url/or/delicious/... /tag/programming" />
> >>    <!-- or even mangle it a bit by still using dc:subject! -->
> >> </item>
> >>
> >> and in the tags RDF...
> >>
> >> <Tag rdf:about="#programming">
> >>    <tagName>Programming</tagName>
> >>    <!-- whatever SKOS properties you'd like to use; it's a subclass of
> >> Concept.
> >>         I'd like a better way of doing names; SKOS doesn't quite work,
> >> as prefLabel and
> >>         altLabel don't capture the intended semantics. -->
> >> </Tag>
> >>
> >>    I'd love some (preferably constructive, but any kind will do!)
> >> feedback from interested parties; blog it, mail me, or mail to the
> >> list. Everything's on the Web:
> >>
> >> RDF/XML: <http://www.holygoat.co.uk/owl/redwood/0.1/tags/>
> >> "hidden" N3 version:
> >> <http://www.holygoat.co.uk/owl/redwood/0.1/tags/tags.n3>
> >>
> >>    The first URL should be persistent (though the ontology contents
> >> are
> >> in flux, of course); the second one should stick around too.
> >>
> >>    Cheers,
> >> -R
> >>
> >>
> >
> >
> > --
> > Seth Russell
> > www.speaktomecatalog.com
> >
> 
> 


-- 

http://dannyayers.com
Received on Friday, 25 March 2005 00:17:33 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 1 March 2016 07:41:45 UTC