Re: [ANN] MOAT

On 21/01/2008, Frederick Giasson <fred@fgiasson.com> wrote:
>
> Hi Danny,
>
>
> > I'm not sure, there may be another opportunity for interop with SKOS.
> >
> > This:
> >
> > <moat:Meaning>
> >    <moat:meaningURI rdf:resource="http://sws.geonames.org/2988507/"/>
> >
> > seems conceptually (!) pretty close to:
> >
> > <skos:Concept rdf:about="http://sws.geonames.org/2988507/">
> >
> > Though I'm not sure what the mapping would look like, given the extra
> > indirection (note that moat:meaningURI is defined as a
> > DatatypeProperty in the ont, but appears as an ObjectProperty in the
> > example).
> >
>
>
> Not certain I agree with you, and I do agree that something look weird
> with meaningURI.
>
>
> Well, there is how I see MOAT and its context. First, you have a literal
> entity that is called a "tag". Tags have a context: the user that used
> it, and is related to the thing it tags. At this moment, I don't see a
> tag as a concept in the sense of a skos:Concept. However, a tag can
> "mean" a concept (a SKOS concept). One tag can be related to one or more
> concepts.
>
> However, right now, moat:meaningURI has a Resource has range. So, I can
> related the meaning of a tag with virtually anything in the World. Does
> this make sense? It really depends on the meaning of a "meaning".
> Otherwise, could the range be a skos:Concept? Yes, I think it could be a
> good idea. But, the system couldn't be able to use dbpedia anymore since
> they are not skos:Concept.
>
> Alex: why the range of meaningURI is a resource? (well, the name make
> sense that the range is a Resource, any resource, but I am not sure the
> name of this property is optimal and unambiguous considering the context
> here).
>
> Why not a moat:concept (range skos:Concept) or something? Thinking aloud
> here.

Personally I would prefer it to be Resource for better
interoperability with actual objects in the field. If people tagged
things to categorise them, as opposed to labelling them, then a
specific range might be relevant, but personally I just label when I
tag so it can be anything that I am labelling the item against. If the
actual URI is a skos:Concept then you can effectively allow for it
still when you make up a query filter on it.

If you make it a moat:concept you add another level of complexity and
reduce the scalability IMO, as one has to augment the URI which
otherwise could have been completely separate from the system. Simple
is good.

There could be a case for having moat:concept with extra annotations
available locally on the concept, but it seems easier to keep the
provenance directly given to the tag by the user context right next to
the meaningURI.

(Hope those three comments made sense together!)

> However, where I think you are right and understood your suggestion,
> after rethinking about it:
>
>
> You are suggesting that a moat:Tag is in relation, with a property like
> moat:concept directly with a skos:Concept instead of a moat:Meaning?
>
> The problem with this, I think, is that you loose the context of the
> meaning of the tag (each meaning is related to a user). Why I do think
> this is important is in a context where you would have 30 different
> meaning for a single tag. What if we do not know where they come from?
> It is where MOAT is really interesting and brings something new (and
> useful!) to tags. Because now, we have a way to manage these tags.
>
> But I still believe that meaningURI could be changed to moat:concept, or
> something similar.
>
> thoughts Alex? Did I miss something?
>
>
> Thanks for the great work!
>
>
> Take care,
>
>
> Fred

Received on Sunday, 20 January 2008 20:17:44 UTC