RE: [Proposal][SKOS-Core] handling top concepts - original use ca se

After skimming through this long dialogue, a few comments from a
thesaurus junkie who doesn't always follow the technicalities of how you
have to do it in RDF:
1. It is absolutely normal for a top concept in one thesaurus to occupy
a lower hierarchical position in another.
2. Similarly, different thesauri usually take different views about
relationships between concepts. For example, some thesauri would give
'cattle' as a NT of 'Organisms'; others would fill in up to 6
hierarchical levels between these concepts. What I am saying is that
relationships between concepts should generally be specified in
association with a particular thesaurus, and not expected to apply
universally. ( Of course, some concepts and relationships do have a
fairly robust universal applicability, but in general you cannot and
should not assume that to be the case.)
3. In the UKAT case cited, yes, I agree that the occupation of a Top
location could be expected to vary between microthesauri, and so
wherever a term/concept is designated as Topterm, that should be
regarded as specific to one or more microthesauri.
4. And should every thesaurus have a "root concept" - the toppest term
of all? Please no, if it means setting up artificial hierarchical
relationships from the genuine top terms. It is true that some system
implementers want and expect it, but it causes confusion for other users
of the same data.
4. And, should  "Top term" be assigned to a term/concept at all? As was
pointed out, to do so is a bit redundant because you can infer it from
the absence of a BT relationship. But in practice I find that people
trying to implement thesauri ( and especially taxonomies) find it very
helpful to be able to identify the top term entries rapidly, directly,
without having to look for the absence of something. System
administrators typically have no previous experience with thesauri and
they want simple things to help them manage the data. So when thesaurus
data are being exchanged or distributed, it is good to give them all the
labels/attributes/properties that will speed up the implementation.
5. As to including a section on 'microthesauri' in guidance notes for
general use, be careful. Not everyone uses this term to mean the same as
UKAT does.

Hopefully my comments support what you were all concluding in any case.
Stella

*****************************************************
Stella Dextre Clarke
Information Consultant
Luke House, West Hendred, Wantage, Oxon, OX12 8RR, UK
Tel: 01235-833-298
Fax: 01235-863-298
SDClarke@LukeHouse.demon.co.uk
*****************************************************



-----Original Message-----
From: public-esw-thes-request@w3.org
[mailto:public-esw-thes-request@w3.org] On Behalf Of Miles, AJ
(Alistair) 
Sent: 04 August 2004 18:11
To: 'Dan Brickley'
Cc: 'public-esw-thes@w3.org'
Subject: RE: [Proposal][SKOS-Core] handling top concepts - original use
ca se



This issue originally came up through collaboration with the UK Archival
Thesaurus team.  UKAT is a thesaurus with several contained
'microthesauri'. UKAT wanted to know how to model microthesauri, which
are not covered by the current SKOS Core guide.  I suggested they model
each microthesaurus as a concept scheme in its own right.  The UKAT
concepts can then be declared as members of both the overarching scheme
and a microthesaurus as well.  Some concepts are top concepts within a
microthesaurus, but not in the overarching scheme - this was the
original use case.  That's when we realised there could be a problem
with skos:TopConcept whenever a concept is a member of more than one
scheme.  

I was thinking about putting a section on 'Microthesauri' in the
'Advanced Features' section of the proposed 'Guide to Using SKOS Core
for Thesauri'
note, explaining how to do it ... what do you reckon?   

(Btw I started sketching a table of contents for that note on the wiki
at 
http://esw.w3.org/topic/SkosCoreGuideToc )

Al.
---
Alistair Miles
Research Associate
CCLRC - Rutherford Appleton Laboratory
Building R1 Room 1.60
Fermi Avenue
Chilton
Didcot
Oxfordshire OX11 0QX
United Kingdom
Email:        a.j.miles@rl.ac.uk
Tel: +44 (0)1235 445440



> -----Original Message-----
> From: Dan Brickley [mailto:danbri@w3.org]
> Sent: 04 August 2004 17:55
> To: Miles, AJ (Alistair)
> Cc: 'public-esw-thes@w3.org'
> Subject: Re: [Proposal][SKOS-Core] handling top concepts
> 
> 
> * Miles, AJ (Alistair)  <A.J.Miles@rl.ac.uk> [2004-08-04 17:27+0100]
> > > Thanks, this identifies a discomfort I've had w/
> interactions between
> > > 'top concept' notion and thesaurus mixing. At heart you're saying
> > > 'top concept' is a relation between a a 
> scheme/dataset/thesaurus and
> > > a concept. Makes sense to me.
> > > 
> > > So would this be:
> > > 
> > > <owl:FunctionalProperty
> > > rdf:about="http:///....../skos/core#hasTopConcept"/>
> > > 
> > > ie. anything that has a skos:hasTopConcept has only one
> such thing?
> > > 
> > 
> > Thanks Dan.
> > 
> > The original idea was that a scheme has several skos:hasTopConcept 
> > properties, pointing to the top level concepts for that
> scheme (i.e. so not
> > functional).
> > 
> > If we made skos:hasTopConcept functional, each scheme would
> have to be
> > defined with a single root concept ... do you think it's
> worth doing it that
> > way?
> 
> Ah, righto. I was reading too much into 'top'.
> 
> Yeah seems more useful to have several, otherwise they'll all just be
> thing/entity/object/resource etc...
> 
> I'm not 100% clear on the use case for this construct, I guess.
> 
> Dan
>

Received on Thursday, 5 August 2004 05:44:38 UTC