semantic web - Named Graphs and Federated SPARQL Endpoints -
i came across working draft sparql 1.1 federation extensions , wondered whether possible using named graphs (not detract usefulness of aforementioned draft).
my understanding of named graphs little hazy, save thing have gleamed reading specs comprises rules around merger, non merger in relation other graphs @ query time. since doesn't satisfy understanding, question follows:
given following query:
select ?something named <http://www.vw.co.uk/models/used> named <http://www.autotrader.co.uk/cars/used> { ... }
is reasonable assume query processor/endpoint or should in context of named graphs following:
check named graph exists locally
if doesn't perform following operation (in case of above query, use second named graph)
get /sparql/?query=encodedquery http/1.1 host: www.autotrader.co.uk user-agent: my-sparql-client/0.1
where encodedquery includes second named graph in from named
clause , where
clause amended accordingly respect graph
clauses (e.g if graph <http://www.vw.co.uk/models/used> {...}
being used).
only if can't perform above, of following:
get /cars/used http/1.1 host: www.autotrader.co.uk
or
load <http://www.autotrader.co.uk/cars/used>
- return appropriate search results.
obviously there might additional considerations around offset
's , limit
's
i remember reading somewhere long time ago in galaxy far far away, default graph of sparql endpoint should named graph according following convention:
for: http://www.vw.co.uk/sparql/ there should named graph of: http://www.vw.co.uk represents default graph , above logic, should possible federate sparql endpoints using named graphs.
the reason ask want start promoting federation across domains in above example, without having wait around standard, making sure won't out of kilter or incompatible else in future.
Comments
Post a Comment