W3C home > Mailing lists > Public > public-esw-thes@w3.org > August 2005

a few questions about collections, schemes, and relations

From: Erik Hennum <ehennum@us.ibm.com>
Date: Thu, 11 Aug 2005 06:45:08 -0700
To: public-esw-thes@w3.org
Message-ID: <OFE81207EE.529B679A-ON8825705A.00446E5C-8825705A.004B8B46@us.ibm.com>




Esteemed SKOSnauts:

I've a few questions; my apologies if these are well-cooked issues.


1. Is it or should it be possible to indicate that a collection is part of
a scheme?

An organization of concepts seems likely to be specific to a scheme,
but I notice that inScheme has a domain of Concept


2.  Is it or should it be possible for a scheme to have a collection as a
top item?

Organizations of concepts seem more likely at the upper levels rather
than lower levels of a scheme, but I notice that hasTopConcept has
a range of Concept.


3.  Is it or should it be possible for a collection subclass to have
ordered and
unordered instances?

For instance, a subclass might indicate the collection principle, as in
RegionalCollection.  In some cases, you might want to order the
regions to reflect the quantity of milk produced in the region but, in
others, list regions in no particular order.

Would it make sense to use the OrderedCollection structure but to offer
properties on the instance (or on the subclass) to indicate whether the
supplied ordering is meaningful?  That would also allow subproperties
that indicate new kinds of ordering.


4.  Is it or should it be possible for concept relationships to be specific
to a scheme?

Hierarchical relationships seem especially likely to be specific
to a scheme.  If an RDF model includes multiple schemes, how
would the relationship indicate its scheme?

For instance, fruit and oranges might be concepts in both produce
and color schemes but only have a narrower relationship in
the produce scheme:

ProduceTaxonomy scheme
    hasTopConcept Fruit
        narrower OrangeFruit
    hasTopConcept Vegetables

Color scheme
    hasTopConcept OrangeColored
        narrower OrangeFruit
    hasTopConcept ManyColored
        narrower Fruit

Being able to indicate the scheme for relationships might address
some of the concerns that motived the scope construct in
TopicMaps (http://www.topicmaps.org/xtm/1.0/#def-scope) -- not
for model interoperability but to build on a developed thought.


Thanks in advance,


Erik Hennum
ehennum@us.ibm.com
Received on Thursday, 11 August 2005 13:46:45 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:38:53 GMT