Quoting Simon Spero <[log in to unmask]>:

)
>
> The MADS/RDF Vocabulary Description
> <http://www.loc.gov/standards/mads/rdf/>states  (§2.3) that
> "ComplexType describes [...] types by aggregating two or
> more Authority and/or two or more Variant records each of a
> SimpleType". Thus, it is not possible for a ComplexType to have a component
> list with only one member.

Simon thanks for finding the relevant section. Now I have a couple of  
other questions:

1) if you have, for example, a simple geographic property, how do you  
know if it is a simple subject heading or something that can only be  
used in a complex subject? A less ambiguous example might be a  
temporalElement, something like "1940-1945." This probably cannot be  
used as a subject heading by itself. Is there a rule in the OWL data  
that indicates this? (I believe that I am referring to the  
free-floating subdivisions that cannot be used alone, but there may be  
other such subdivision types in LCSH as well.)

2) Isn't it the case that a variant heading could be composed of a  
variant simple type in position 1 followed by authoritative simple  
types in the subsequent positions? (and perhaps even an authoritative  
entry in position one and a variant entry in a subsequent position?)

3) What does the "and/or" indicate in the above section? What does it  
mean to have two or more authoritative AND two or more variant records  
in a complex record?

I suspect that there is something buried deep in the OWL definitions  
that may clear this up, but like my first question, I haven't found it.

kc


>
> However, this same quotation is a good example of some of the fundamental
> confusions that render the ontology as a whole inconsistent.
>
>    - A restriction is given in the text but not stated in the ontology (that
>    component list must be a List with at least two members, and that each
>    member must be a SimpleType and either an Authority or a Variant)l. This
>     restriction can be expressed in OWL.
>    - Authority and Variant  are disjoint (one is a subclass of skos:Concept;
>    the other a subclass of skosxl:Label). It is not clear what  
> intuitive class
>    corresponds to "Concept or Label".
>    - The examples that use a componentList contain members that are
>    SimpleTypes but which are not marked as  either being instances  
> Authority or
>    of Variant.
>    - It is not intuitively clear what is denoted by the aggregation of a
>    Label with a Concept.
>    - The style of specification makes it difficult to express rules for the
>    order of subdivisions that are part of the SCM, making it less useful to
>    organizations following LC cataloging and policy standards (for  
> example, LC)
>
> The specification fails to capture any of the semantics of coordinated
> headings. There are some semantics that differ  depending on whether one is
> applying a faceted or subdivision interpretation to these coordinate
> headings, but much of the semantics is common to both, and none are
> captured.
>
> Simon
> p.s.
> I would avoid using the word
> *property<http://philpapers.org/browse/property-nominalism>
> * here, as it has pre-existing Ontological meaning. §2.2 of MADS/RDF shows
> an  eliminativist approach to Properties that would be significant were it
> not clear from the text that it it is un-intentional as well as
> un-intensional.
>



-- 
Karen Coyle
[log in to unmask] http://kcoyle.net
ph: 1-510-540-7596
m: 1-510-435-8234
skype: kcoylenet