Re: [SKOS] ISSUE 77 and ISSUE-40: proposal for postcoordination

Hi all,

Alasdair wrote:

 > Aida wrote:
 >>
 >>> 1. Indexing: How do you encode the statement "Person <P> indexed
 >>> resource <R> with concepts <C1> and <C2>"?
 >> IMHO this has nothing to do with SKOS or with vocabulary as such.
 >> In all systems that I know of, this information is normally encoded
 >> in meta-metadata - or administrative metadata. It is not part of
 >> vocabulary or resource. These meta-metadata hardly have any value
 >> outside local system scenario and many cataloguing agencies never
 >> export it or include it in the process of information exchange. You
 >> have to come up with some good argument why this information would
 >> be relevant for resource discovery. Authority of metadata is usually
 >>  established through institutions not through individuals.
 >
 > I agree with this point of view: keep SKOS for capturing the generic
 > features of the KOS and then allow systems to apply the KOS as they
 > see fit. However, indications on how this can be achieved and whether 
 > any additional machinery are required should be considered, but in the
 > end, the document which publishes the KOS should be kept focused on
 > that one topic so that it can be easily reused.

Well, then we just don't agree in this point. I think Social Tagging and 
Folksonomies show that you cannot divide an indexing vocabulary and its 
usage (who indexed what in which context). Looks like we hit a current 
research frontier of information science ;-)


Alasdair wrote:

 >> As mapping can be done as direct mapping or through intermediary
 >> vocabulary (pivot/spine vocabulary) <W>. So this may be published as
 >> vocabulary crosswalks <X => W> and <Y =>W>.
 >
 > I don't think that approach works. In some cases, you will still need
 > some way of saying that <A> exactMatch (<B> AND <C>). This is an
 > important area which is lacking from the current SKOS reference.

That's the point. And I even think that these case are not just the 
exception!

What do you think about the following solution:

The intersection (<B> AND <C>) is a subset of <B> and a subset of <C>, 
right? So you can say

<B> skos:narrower (<B> AND <C>) .
<C> skos:narrower (<B> AND <C>) .

The mapping

<A> exactMatch (<B> AND <C>)

can then already be expressed with current SKOS:

<A> skos:exactMatch [
   rdf:type skos:Concept ;
   skos:broader <B> ;
   skos:broader <C>
] .

Furthermore the union (<B> OR <C>) is a superset of <B> and a superset 
of <C>. The mapping

<A> exactMatch (<B> OR <C>)

can then already be expressed with current SKOS:

<A> skos:exactMatch [
   rdf:type skos:Concept ;
   skos:narrower <B> ;
   skos:narrower <C>
] .


Isn't that pretty? :-)


Aida wrote:

> Coming back to your real problem.
> For pre-coordination for the moment I don't have a definitive 
> proposition, but for post-coordination, why don't you use rdf:Bag and 
> rdf:Alt [1]? Has it been proposed before?

[...]

> So what do you think this proposal is usable for you? I do believe it is 
> simpler, and appropriately re-uses existing RDF solutions (I think the 
> examples given in [1] are really close to what you want to achieve)

Well, I'll have a look. Personally I find rdf:Bag and rdf:Alt almost as 
ugly as reification but maybe it can help.

> Notice we MAY want to assign semantics for this proposal pattern, e.g. 
> saying that all the members of the bag are the skos:subject of the 
> document. But I'm not sure this actually fits the intended semantics of 
> post-coordination. If a document is about a combination of concept in a 
> post-coordination-based search system, it is not supposed to be "fully" 
> about each concept.

Good point. On the other hand that's an argument to define more precise 
semantic instead of just using rdf:Bag/rdf:Alt.

Greetings,
Jakob

-- 
Jakob Voß <jakob.voss@gbv.de>, skype: nichtich
Verbundzentrale des GBV (VZG) / Common Library Network
Platz der Goettinger Sieben 1, 37073 Göttingen, Germany
+49 (0)551 39-10242, http://www.gbv.de

Received on Tuesday, 18 March 2008 11:27:05 UTC