[Esip-discovery] Linking datasets and services

[Lynnes, Christopher S. (GSFC-6102)]

So Ruth and Brian have probably already suggested this but I didn't grasp the significance.  So let me see if I can describe a linking framework in a fairly straightforward manner to include the datasets that a given service works with by including <link> elements that reference canonical dataset landing pages, datacasts or even opensearch queries corresponding to datasets that work with or are referenced by that service. Basically, we could use rel="search" type="application/opensearchdescription+xml", which would point to an Open Search Description Document that in turn describes a query or pseudo-query that returns an Atom response of data that are compatible with the service.

This could be used in several different ways:
(1) A granule-level OSDD, i.e., with the dataset embedded in the Url template. We may not actually need to know what the dataset IS, just that if we do a query with that template we are guaranteed to get granules that work with the service

(2) A dataset-level OSDD, but with the URL template including some parameter that constrains the response to datasets that work with that service, e.g., <Url template="http://some.org/q?fmt=netcdf&processingLevel=3&keywords={searchTerms}". This would be for the case of a service that only works with level 3 netCDF data, for example. 

(3) A pseudo-OSDD whose Url template is a static URL pointing to a datacast Atom file, either dataset-level or granule-level.  OK, "search" as a rel is a stretch here, but who's going to know?

???
--
Dr. Christopher Lynnes, NASA/GSFC, ph: 301-614-5185