atom feeds as resource maps

[Ed Summers]

Does the constraint in the Data Model that a Resource Map MUST
describe a single Aggregation [1] prevent someone from using an Atom
Feed (containing multiple atom:entry elements) as a OAI-ORE Resource
Map?

I was just reviewing a paper that Michael Witt is writing for Library
Technology Reports about OAI-ORE (really good stuff) and was struck by
this constraint for the first time.

//Ed

[1] http://www.openarchives.org/ore/1.0/datamodel#Resource_Map

--
You received this message because you are subscribed to the Google Groups "OAI-ORE" group.
To post to this group, send email to oai...@googlegroups.com.
To unsubscribe from this group, send email to oai-ore+u...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/oai-ore?hl=en.

[Carl Lagoze]

Ed,

You can definitely package multiple atom entries, each of which corresponds to a resource map, into a single atom feed. We discussed this possibility and the advantages of it when we moved from the earlier implementation in which there was a correspondence between resource maps and atom feeds.

Carl

[Ed Summers]

Thanks Carl--that's good news. So is the language [1] in the Data
Model document that says a Resource Map MUST describe a single
Aggregation out of date, or in need of adjustment?

[Carl Lagoze]

Definitely not. I think you are confusing the concrete atom XML document with the abstract underlying model. Note that the notion of the resource map is tied to the web architecture notion of a resource. Each resource map has a unique URI and must describe a single aggregation, which has another unique URI. That is different from an atom entry, which is an XML fragment that we are employing to serialize a single resource map. That fragment can then be wrapped into a single atom feed document, which can contain multiple atom entry XML fragments. Each of those serializes a single resource map and each has a unique URI (and is therefore a single resource).

I hope this makes sense.

[Ed Summers]

OK, so the answer to my original (admittedly vague) question:

"""
Does the constraint in the Data Model that a Resource Map MUST
describe a single Aggregation [1] prevent someone from using an Atom

*Feed* (containing multiple atom:entry elements) as a OAI-ORE Resource
Map?
"""

is Yes.

In terms of webarch: a web resource that has an application/atom+xml
representation, which contains an atom:feed root element and multiple
child atom:entry elements can not a be considered a Resource Map. The
URI for the feed can't be the URI for the Resource Map. The atom feed
just references Resource Maps (w/ link rel="self" in the atom:entry
elements).

The requirement instead is that an application/atom+xml representation
for a Resource Map should have an atom:entry root element?

Is that about right?

//Ed

[Herbert Van de Sompel]

Ed

I think that Figure1 of http://www.openarchives.org/ore/1.0/atom gives a good perspective of how an ORE Resource Map relates to an Atom entry.

Note that in an early version of the Atom serialization for ORE, a Resource Map was mapped onto an Atom feed with each Atom entry holding a single Aggregated Resource. This was fundamentally revised (by popular demand) to map a Resource Map to an Atom entry.

Cheers

Herbert

Sent from my iPad

[MichaelNelson]

Hi Ed,

Just to amplify some of the points that Carl made:

- An Atom feed has no ORE semantics, but it obviously makes
a convenient container for bundling up lots of Resource Maps
(i.e., Atom entries). All the ORE semantics are at the level of
Atom entries.

- An earlier version of ORE had the Resource Map at the level
of an Atom feed, but we felt that caused more complications than
it solved so we went with the current mapping of Resource Map
to Atom entry.

- The data model language is there to restrict someone from
saying they have a single Resource Map that describes multiple
Aggregations. While it is possible to have multiple Resource Maps
describing a single Aggregation, either through different formats
(e.g., RDF/XML vs. Atom) or through "non-authoritative resource maps"
([1], possible but not recommended), we did not want a single
Resource Map to describe multiple Aggregations.

regards,

Michael

1. http://www.openarchives.org/ore/1.0/datamodel#Resource_Map

[MichaelNelson]

>
> The requirement instead is that an application/atom+xml representation
> for a Resource Map should have an atom:entry root element?
>
> Is that about right?
>

yes.

and as per RFC 5023, the MIME type could more accurately be:
"application/atom+xml;type=entry"

regards,

Michael

[Ed Summers]

Great, thanks for the feedback all.

//Ed