Re: Open Library and RDF

Tom,

I didn't mean to imply that Creator *should* be modeled as a class name.
It's more of a comment about naming properties more carefully to avoid
this confusion.

Jeff

Thomas Baker <tbaker@tbaker.de> wrote:

Hi Jeff,

On Mon, Jul 19, 2010 at 06:02:59PM -0400, Young,Jeff (OR) wrote:
> IMO, the term "Creator" implies a class name. In contrast, "is the
> creator of" or "was created by" implies a property. I promise not to
> whine if "creator" is defined as a property, but only if its range is a
> class named "Creator". ;-)

Hmm, I would argue that it is unnecessary and perhaps even
counterproductive to create a class named "Creator" :-)

The Dublin Core property dcterms:creator, for example, has
a range of dcterms:Agent, not Creator.

What if a given member of a Creator class were the Subject
of a book, or its Translator or Illustrator?  Triples
saying that that person is _essentially_ of type "Creator"
(or Subject or Translator or Illustrator) could muddy the
waters: an Illustrator has translated this and been the
subject of that...

To my way of thinking, in the example above, it is better
simply to use a property to establish a creation relationship
between a given person and given resource.

I'm not entirely up-to-date on the RDA-in-RDF discussions but
I have the impression that this still is an issue there too.

Are there thought to be things that are _essentially_ members
of a class Manifestation, and others _essentially_ of a class
Expression, such that it would a logical contradiction in
terms of the RDA data model if a given resource is asserted
to be a member of both the class Expression and the class
Manifestion?  If so, can we be confident that RDA-trained
(or non-RDA-trained) catalogers will use the classes so
consistently that contradiction need not be feared?  Is it
not enough to describe manifestions in a certain way (i.e.,
with certain properties) and expressions in another?

More generally, if "A hasFriend B", is B essentially a member
of the class "Friends"?  "Friend" with respect to whom?
("Creator" of what? "Manifestation" of what?)

The point of modeling style is that it can be expressive
enough to let the constellation of relations (RDF properties),
within which a resource is embedded, define that resource --
in effect to describe things more with verbs than with nouns.

Maybe we need a Strunk-like "Elements of RDF Style":
"Omit needless classes!" "Vigorous modeling is concise!"
"Do not explain too much!"... :-)

Tom

--
Thomas Baker <tbaker@tbaker.de>

Received on Saturday, 14 August 2010 19:40:43 UTC