W3C home > Mailing lists > Public > public-rdf-dawg@w3.org > January to March 2011

Re: Test cases & entailment

From: Axel Polleres <axel.polleres@deri.org>
Date: Tue, 15 Feb 2011 14:35:41 +0000
Cc: "SPARQL Working Group" <public-rdf-dawg@w3.org>
Message-Id: <4EA61A0A-CD46-46B5-A334-4D1862DF1EB6@deri.org>
To: Birte Glimm <birte.glimm@comlab.ox.ac.uk>
Hi Birte, 

Answers inline.

On 8 Feb 2011, at 20:15, Birte Glimm wrote:

> Hi all,
> I am just working my way through the test cases with our SPARQL 1.1
> implementation (ARQ+HermiT to do OWL Direct Semantics entailment). One
> issue it that it would be very useful if we could mark test cases in
> general with the entailment regimes that can be used. In many cases,
> the query executed with and without entailment yields the same result,
> e.g., when testing aggregates or bind the BGP matching itself is
> usually trivial, and it would be very useful if we can then mark the
> test somehow as such. I.e., I would like to annotate tests as being ok
> also with an entailment regime if the use of entailment regime makes
> no difference to the results.
> The only problematic regime is OWL Direct Semantics, since the graphs
> have to be OWL DL conform, which requires an extra triple for the
> ontology header and declarations that identify a property as object or
> data property. This usually does not influence the results and we
> could make tests suitable also for OWL DL whenever that does not
> influence the results. Any support or objections for that? I would
> volunteer to extend the data and annotate the tests accordingly. As a
> result, we would have several of the normal tests annotated as also
> suitable for some ent. regime and the tests in the entailment folder
> as those that really require entailment to get the right results.
> 
> I am also wondering why the tests in the entailment folder use a
> slightly different marking than what is described in the
> http://www.w3.org/2009/sparql/docs/tests/README.html
> The above uses:
> :rdf01 rdf:type mf:QueryEvaluationTest ;
>     mf:name    "RDF inference test" ;
>     dawgt:approval dawgt:NotClassified ;
>     mf:action
>          [ qt:query  <rdf01.rq> ;
>            qt:data   <rdf01.ttl> ;
>            sd:entailmentRegime ent:RDF ] ;
>     mf:result  <rdf01.srx>
>     .

As from our mail exchange, http://lists.w3.org/Archives/Public/public-rdf-dawg/2010JulSep/0122.html
I think I simply picked that one (that's why I put it like that in the README.html)

> 
> The manifest in the entailment folder uses:
> :rdf01 rdf:type mf:QueryEvaluationTest ;
>     mf:name    "RDF inference test" ;
>     dawgt:approval dawgt:NotClassified ;
>     sd:entailmentRegime ent:RDF ;
>     mf:action
>          [ qt:query  <rdf01.rq> ;
>            qt:data   <rdf01.ttl> ] ;
>     mf:result  <rdf01.srx>
>     .
> 
> Which one did we decide on?
> 
> In order to annotate existing non-entailment tests as also applicable
> under an entailment regime, how about annotating them with a list of
> applicable regimes? For example,
> 
> :rdf01 rdf:type mf:QueryEvaluationTest ;
>     mf:name    "RDF inference test" ;
>     dawgt:approval dawgt:NotClassified ;
>     sd:entailmentRegime (
>         ent:RDF
>         ent:RDFS
>     ) ;
>     mf:action
>          [ qt:query  <rdf01.rq> ;
>            qt:data   <rdf01.ttl> ] ;
>     mf:result  <rdf01.srx>
>     .

We could do the same though with either
 a) replicating the test cases with the same input/output for different ent regimes.
 b) just having multiple values for the sd:entailmentRegime property (that would need some adaption in http://www.w3.org/2009/sparql/docs/tests/README.html#entailevaltests

a) requires no changes, 
if b) is preferred, I would just add one sentence:
"In the absence of the sd:entailmentRegime property, the entailment regime defaults to http://www.w3.org/ns/entailment/Simple, multiple values for the sd:entailmentRegime property indicate that the test case is expected to provide the same result for any of those entailment regimes."

Does that work for you?

Axel


> This would then indicate that the test can be executed with RDF and
> RDFS entailment?
> It might then even be clearer to not abuse the sd:entailmentRegime
> property and instead use something like mf:applicableWith.
> 
> Birte
> 
> 
> 
> --
> Dr. Birte Glimm, Room 309
> Computing Laboratory
> Parks Road
> Oxford
> OX1 3QD
> United Kingdom
> +44 (0)1865 283520
> 
> 
Received on Tuesday, 15 February 2011 14:37:18 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 16:15:45 GMT