Comments on SKOS Core Guide & Quick Guide

Hello,

Guus Schreiber asked me to comment on the SKOS Core Guide [1] and the
Quick Guide to Publishing a Thesaurus on the Semantic Web [2],
as I am also interested in SW representations of thesauri. My apologies 
if these issues were already discussed. Below I start with [1].

First of all, I think this is all very useful material and this document
makes significant contributions to previous versions of SKOS I've seen.
Especially the inclusion of superproperties publicNote and privateNote
really adds useful semantics. It's also well-written.

A first question I have when reading the abstract is that I'm not sure
what exactly the difference is between a conceptualisation and
linguistically oriented knowledge organisation systems. Why is this
distinction important?

The introduction explains how examples, rules and constraints are
depicted throughout the document. For a constraint, I'm not sure I
exactly understand what it means. Is it to give information on the
intended use of a vocabulary item? Also, the intro might also explain
what the acronym SKOS stands for.

Concerning the section "lexical labelling": maybe it's useful to explain
what a language tag is, or is the intended audience probably aware of
it? The section also gives the advice to use rdfs:label if there is only
one label for a concept. Is there a specific reason not to use
skos:prefLabel instead?

Some remarks concerning Symbolic Labelling (skos:prefSymbol/altSymbol)
and Depictions (foaf:depiction). From the text it is not completely 
clear to me what the difference is. Is a depiction broader than a 
symbol? (E.g. the concept "love" has a heart as symbol, but a kissing 
couple is a depiction and not a symbol.)

I was wondering about the difference between skos:scopeNote and
skos:definition (and also editorialNote/changeNote). Thesauri in the ISO 
2788 format only have a scope note (i.e. the scope note is the 
definition). Their difference may become clearer if an example 
containing both a scope note and a definition is included. Also, people 
migrating from an ISO thesaurus need to be aware that their ScopeNotes 
should probably be migrated to skos:definition.

A general comment is that I'm curious about the relationship of SKOS to
standards (e.g. for thesauri). For example, was ISO 2788 an inspiration?
Why have some things from ISO gone in SKOS Core and others not (e.g.
compound terms)? Reference is made to BS8723. Is this a standard of the 
British Standards group (http://www.bsi-global.com)? I haven't been able 
to find a standard with this code on that website.

Would it be possible to have a separate Facet class which is not a
subclass of skos:Concept? A disadvantage of making facets a subclass of
concept is that they can have skos:broader and skos:related relations.
As far as I understood a facet is some sort of "grouper", not in itself
a type of concept.

What is the motivation to use a separate Collection vocabulary instead 
of the standard RDF vocabulary? Does that have something to do with the 
special rules that are introduced for them (collectable property rule)?

Small typos:
- "In practise," (section "Concepts in Multiple Schemes")
- "no mote than one" (section "Annotation Constraints")

When I read the section "advanced annotations" I was a bit confused by 
the term "annotation properties", because of OWL's "annotation 
properties", maybe another name is more neutral?

Lastly, I have two questions concerning the use of the rdfs:isDefinedBy 
property in SKOS:

- If a skos:definition is used in conjunction with the "Document 
Reference Usage Style", is this equivalent to the rdfs:isDefineBy property?

- I don't understand the "defining scheme rule"
   (x isDefinedBy y -> x inScheme y) for two reasons:

   (1)     The RDFS spec says that isDefinedBy is used to point to a
     resource y (possibly not web-retrievable) that defines x. What
     is the relation to a property defining membership of a Concept
     to a ConceptScheme? I don't think a ConceptScheme itself really
     "defines" a concept in the same way a piece of text or a
     document does.

   (2)    How does a general (x isDefinedBy y) statement in a random RDF
     graph imply that x is a skos:Concept and y is a
     skos:ConceptScheme? Should the implication be the other way
     around?

--------------------------------------------------------------------

Below follow some remarks concerning the Quick Guide to Publishing a
Thesaurus on the Semantic Web [2].

It's a good idea to write a Quick Guide to Publishing a Thesaurus on the 
Semantic Web [2]. I think it will be a very welcome resource to 
thesaurus people just getting into the SW.

The example is a good way to start, because the basic idea becomes clear 
quickly. However, I think for many readers some more information on SKOS 
is needed before they can understand the example. E.g. some will have 
difficulty seeing that a Term (as appears in ISO 2788 thesauri) is 
equivalent to a skos:Concept.

A question on the scope: is the idea of the Guide also to explain how to 
convert a thesaurus to OWL, or is this outside the scope. Of course 
there are lots of additional issues such as multilingual thesauri, 
thesauri that do not fit SKOS, etc. A separate section which just names 
the issues and points to additional material would be great.

I don't know if this is outside of the scope of the guide, but in [3] I
and some collegues separated the conversion of a thesaurus into 
different sequential steps. It may be material for the "Further Reading" 
section.

--------------------------------------------------------------------

Hope these are useful comments. If I can help any further I will
gladly do so!

Kind regards,
Mark van Assem.

---
[1] http://www.w3.org/2004/02/skos/core/guide/2004-11-25.html
[2] http://www.w3.org/2004/03/thes-tf/primer/2004-11-17.html
[3] http://www.cs.vu.nl/~mark/papers/Assem04a.pdf

-- 
  Mark F.J. van Assem - Vrije Universiteit Amsterdam
        mark@cs.vu.nl - http://www.cs.vu.nl/~mark

Received on Wednesday, 12 January 2005 01:18:31 UTC