I feel it necessary to discuss the best practices of separation between template definitions and reference data proper in an endpoint.
There are three options available:
- No separation at all, complex SPARQL queries can extract data based on type declarations.
- Separate graphs at one query interface URL.
- Separate query interfaces.
It seems that majority of existing open sandboxes contain parts of template definitions together with reference data, but I do not know whether these are remains of some experiments, or some people think it is a proper practice.
Personally I'd prefer to see complete separation with template definitions at their own query interface, like it is done for the main iRING sandbox.
Template and RD endpoint separation
-
- Posts: 283
- Joined: Sun Jan 22, 2012 10:02 pm
Re: Template and RD endpoint separation
Hi Victor,
Could you give me some simple examples of the three options?
That would help me understand it.
Regards,
Hans
Could you give me some simple examples of the three options?
That would help me understand it.
Regards,
Hans
Re: Template and RD endpoint separation
Sorry, I wasn't notified of your reply in time somehow.
1. No separation at all:
Look at one of sandboxes:
Query:
Query:
Everything goes together - part 2 type instances, template roles. Some reasoning on client's side required.
2. Separate graphs:
Look at http://jord-dev.org/endpoint/ , the last of Query List, looking for R-number
Only part of triple server is searched, graph http://irm.dnv.com/ontologies/iring.map with bridge from R to RDS numbers.
3. Complete separation. That's what we know well:
http://posccaesar.org/endpoint/sparql - only proper Part 2 instances.
- only templates
1. No separation at all:
Look at one of sandboxes:
Code: Select all
http://www.iringsandbox.org/repositories/SandboxPt8/query
Code: Select all
PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
SELECT *
WHERE
{?uri rdf:type <http://dm.rdlfacade.org/data#ClassOfEvent>}
LIMIT 1000
Code: Select all
SELECT *
WHERE
{<http://tpl.rdlfacade.org/data#R98830569416> ?p ?o}
ORDER BY ?label
LIMIT 1000
2. Separate graphs:
Look at http://jord-dev.org/endpoint/ , the last of Query List, looking for R-number
Code: Select all
PREFIX RDL: <http://jord-dev.org/rdl/>
PREFIX afn: <http://jena.hpl.hp.com/ARQ/function#>
SELECT * {
GRAPH <http://irm.dnv.com/ontologies/iring.map> {
?R RDL:rdsWipEquivalent ?RDS .
FILTER (afn:localname(?R)= "R6565765")
}
}
3. Complete separation. That's what we know well:
http://posccaesar.org/endpoint/sparql - only proper Part 2 instances.
Code: Select all
http://www.iringsandbox.org/repositories/tempInitSet/query