I said:

> What if we adopted a vocabulary that used only the "content" types
from
> CLDT? i.e. those "corresponding to" (but not the same as!) the DCMI
Type
> Vocabulary terms (probably excluding a Collection of Collections and a
> Collection of Services) :

Diane said:

> I think this would be a terrible mistake. If you want to convey the
notion
> of a "Collection of Collections," a "Collection of Images" or a
"Collection
> of Services" for goodness sake use those terms unambiguously, or rely
on simple
> repetition of DCMI Type terms to do the job. Re-using DC Type terms
and expect
> the schemas or definitions to keep them differentiated strikes me as a
very bad
> idea indeed.

OK. The suggestion was that (as a sort of lowest common denominator) we
could adopt a set of Collection Types for which there is a one-to-one
relationship to the DCMI Types. Sarah's account of the deliberations at
IMLS suggest they have settled on something like this too.

i.e. the DCMI Type vocabulary allows us to say:

This resource is a text ("... a resource whose content is primarily
words for reading. For example - books, letters, dissertations,
poems...")
[etc]

And the Collection Type vocabulary would allow us to say:

This resource is a collection of texts (resources "...whose content is
primarily words for reading. For example - books, letters,
dissertations, poems...")
[etc]

I think these are two _different_ statements so we would need different
sets of types/classes to express them, i.e. it wouldn't be a straight
reuse of the DCMI Type Vocabulary, but it is based on the categories
defined by the DCMI Type Vocabulary.

To ensure that the different meaning is clear from the label even when
it is not "qualified" by the name of the encoding scheme, we
could/should assign the individual terms in the second set of Types
different human-readable labels from the labels of the terms in the
first set. And similarly we can try to ensure that the URIs are such
that the XML QName forms are more easily distinguished by the human
reader too.

Something like

URI = http://example.org/coll/type/CollectionText (or
.../TextCollection)
QName = collType:CollectionText (or collType:TextCollection)
Label = Collection of Texts
Definition: This resource is a collection of texts, where "text" is
defined by the corresponding term in the DCMI Type Vocabulary
http://purl.org/dc/dcmitype/Text

URI = http://example.org/coll/type/CollectionImage (or
.../ImageCollection)
QName = collType:CollectionImage  (or collType:ImageCollection)
Label = Collection of Images
Definition: This resource is a collection of images, where "image" is
defined by the corresponding term in the DCMI Type Vocabulary
http://purl.org/dc/dcmitype/Image

etc

I realise this then has elements of the "Collection.*" composite about
it, though :-(

Pete

P.S. When I said "I can't argue that type vocabularies are difficult", I
meant "I can't argue with the statement that constructing type
vocabularies is difficult"...!