I confused some of you with my OIA reference so let me elaborate.
I wasn't thinking at all of SRW as a replacement for the OAI protocol.
I see the OAI protocol as continuing to serve as the protocol for
harvesting.
As Ralph so enthusiastically points out, the natural place for SRW in
a metadata-harvesting environment is to access the harvested metadata.
However, our (very simplistic) idea isn't that, not initially.
Our idea is this: Assume there is an OIA data provider, accessible
(only) via the OIA protocol, and an SRW client wants to access it. So
we build a gateway. Why the SRW client wants to access the raw OIA
data, we need to develop a little bit further; but not to do any
meaningful searching. Perhaps to retrieve specific records by
identifier. Or to retrieve a list of identifiers, a list of supported
formats, even a list of sets, and these would be via SRW explain,
assuming that we develop the explain schema to support this, which
further assumes that this is a useful thing to do. Anyway, we have OAI
data here (Dublin Core records), accessible only via the OIA protocol.
So that's why we've come up with this idea.
--Ray
|