[OAI-implementers] repository auto-discovery

[Dr R. Sanderson]

On Sun, 19 Nov 2006, Robert Forkel wrote:

> On 11/19/06, Herbert Van de Sompel <herbertv at lanl.gov> wrote:
>
>> (2) A facet of the problem domain that unAPI covers has been explored in
>> other contexts.  A generic formulation of that problem could be (if I
>> dare) "How to obtain (in an interoperable manner) a manifest of all
>> available Representations of a Resource" (to use W3C Web architecture
>> language).
>
> well, i'm aiming at something much lower: namely, how to get the baseUrl of
> an OAI PMH data provider? and it seems particularly embarassing, that i have
> no standard way to advertise my own service to people (including robots)
> surfing my own pages.

I think this depends a lot on what you're considering as the entry point 
to the service.

For example, it's trivial to link to the identify page and use XSLT to 
turn it into a user interface, as demonstrated by OCLC's repositories, 
amongst others.  Tom Habing has demonstrated the usefulness of simply 
linking to the identify page in his google search discovery technique 
for repositories.  So you have an easy way to advertise your repository 
as a whole. (See also OpenSearch's OSD, SRU's Explain)

However, by your reference to unAPI, I suspect you want a deep link into 
the repository for a particular item (like Ed's example)?  At which 
point I think the unAPI analogue breaks down -- unAPI gets you this item 
that you're looking at on the screen right now, whereas OAI is more 
geared towards getting you all of the items in the repository that 
you're after (by set, by date, by id, etc).  So in this case, why not 
simply use unAPI or OpenURL?

What I wouldn't like to see is further division of the environment into 
16 different autodiscovery links for any item. Already there could very 
easily be:  OpenURL (the *standard* for this sort of thing! ...), unAPI 
(... however there is much to be said for 1 page specs),  OpenSearch, 
OAI, SRU, AssetAction, etc etc. Plus I'm sure that the microformat 
people will invent their own sliiiiightly different versions of all of 
the above, and so forth.  Semantics are only useful when they're shared.


Rob Sanderson