Just to pick up on what I said the other day

> I admit I do have a slightly niggling doubt about whether
> we're on completely
> firm ground treating what RSLP CD calls a Location as a
> Service, but I think it
> does make sense - certainly for the case of the digital
> collection, I don't understand what else a Location can be.

I'm really not sure whether there is a problem here or not, but the
dc:publisher discussion has set me thinking about it and, FWIW, I've
tried to sketch some diagrams to illustrate the distinctions between
these two possible models.

The "core" data model underpinning the current DC CD AP is essentially
that

- a Collection is made available by/at one or more Services/Locations

i.e. it treats a Physical Location (a physical repository) and a Network
Service (a Web site, a Z-target, an OAI-PMH repository etc etc) as
sub-types of the same Location/Service entity type so:

A physical collection (a library collection) is made available at/by a
physical location-service (a library service)

http://homes.ukoln.ac.uk/~lispj/dc-cd/model/model/Slide1.jpg

A digital collection (an image library) is made available by a network
service (a web site)

http://homes.ukoln.ac.uk/~lispj/dc-cd/model/model/Slide2.jpg

and so on

http://homes.ukoln.ac.uk/~lispj/dc-cd/model/model/Slide3.jpg
http://homes.ukoln.ac.uk/~lispj/dc-cd/model/model/Slide4.jpg
http://homes.ukoln.ac.uk/~lispj/dc-cd/model/model/Slide5.jpg

What I'm wondering is whether conflating Service and Location in this
way is really the right way to model this, and whether we need to say
that

- a Collection is located at one or more Locations, and each Location is
accessible via one or more Services

So the corresponding examples now look like:

A physical collection (a library collection) is located at a physical
location (a library) and access is provided by a physical
location/service (a library service)

http://homes.ukoln.ac.uk/~lispj/dc-cd/model/model/Slide6.jpg

A digital collection (an image library) is located at a digital location
(a database/file store) and made available by a network service (a web
site)

http://homes.ukoln.ac.uk/~lispj/dc-cd/model/model/Slide7.jpg

and so on

http://homes.ukoln.ac.uk/~lispj/dc-cd/model/model/Slide8.jpg
http://homes.ukoln.ac.uk/~lispj/dc-cd/model/model/Slide9.jpg
http://homes.ukoln.ac.uk/~lispj/dc-cd/model/model/Slide10.jpg

I think we've tended to equate Service and Location because in the
digital case, the "user" usually only sees (the access point of) a
Service, and it is of little interest to that user where the data
delivered via that Service is actually stored: in effect the Service is
the Location.

For the physical case, I think the discussion about dc:publisher
highlights that treating Service and Location as a single entity _may_
be slightly flawed.

For the DC CD AP we are concerned with the properties of the Collection,
but the choice of model determines whether we wish to describe

(a) the relationship between a Collection and a Service

(which is the basis we have been working on to date, I think) or

(b) the relationship between a Collection and a Location (which is in
turn made accessible via a Service)

I think we need to try to examine the pros and cons of the two
approaches.

Pete