Re: ISSUE-28 (equivalent blank nodes): Clarify whether blank nodes are equivalent based on their identifier [R2RML]

* Richard Cyganiak <richard@cyganiak.de> [2011-03-22 11:13+0000]
> On 21 Mar 2011, at 15:48, David McNeil wrote:
> > Alex - This test looks useful. This raises the question of how to verify the test results for TriplesMaps that produce blank nodes. It seems to me that a test harness for executing the test cases will need to ignore the blank node identifiers used (since an implementation is free to generate any blank node identifiers).
> 
> I'd think so.
> 
> > Do we expect that a test harness will be created by the working group?
> 
> I can't answer this, but FWIW Jena has a method for checking “graph isomorphism”. It checks whether graphs A and B are isomorphic, that is, there is a 1:1 mapping from blank nodes in A to blank nodes in B that transforms graph A into graph B. I've used it in the past for exactly this kind of testing. I'd expect that similar algorithms are found in other RDF APIs too (but certainly not in all).

For historical precedent, the SPARQL 1.0 test suite depended upon
graph isomorphism with respect to bnode relabeling; it is likely that
most of the implementations in
http://www.w3.org/2001/sw/DataAccess/impl-report-ql relied on
automated isomorphism tests.

Our implementations probably don't require isomorphism equivalence
apart other than for testing, so it may well be most useful to create
a service which has the reference results and compares to the expected
results. Alexandre and I could stick something into
http://this-db-really.does-not-exist.org/ .


> Best,
> Richard
> 
> 
> > 
> > -David
> > 
> > On Mon, Mar 14, 2011 at 1:09 PM, Alexander De Leon <me@alexdeleon.name> wrote:
> > Created a basic test case for this:  http://www.w3.org/2001/sw/rdb2rdf/wiki/R2RML_Test_Cases_v1#DB26
> > 
> > Alex
> > 
> > On 07/03/2011, at 17:02, RDB2RDF Working Group Issue Tracker wrote:
> > 
> > >
> > > ISSUE-28 (equivalent blank nodes): Clarify whether blank nodes are equivalent based on their identifier [R2RML]
> 
> 

-- 
-ericP

Received on Tuesday, 22 March 2011 15:40:16 UTC