Re: First version of a test suite for RDF*

Holger,

On 23/11/2020 22:23, Holger Knublauch wrote:

>
> On 11/24/2020 7:06 AM, Pierre-Antoine Champin wrote:
>>
>> On 23/11/2020 14:39, Ghislain ATEMEZING wrote:
>>
>>> Hi Pierre-Antoine,
>>> Many thanks for this great input.
>>> Just to be sure I understand the coverage of the test suite: Are we 
>>> going to add also RDFS entailment regime?
>> The test-suite as it is is parial, just a first shot. And yes, the 
>> idea is to add RDFS.
>>> In general, how to be sure we take into account all the relevant 
>>> cases? #justAsking.
>>
>> Just like any test-suite, you can never be sure...
>>
>> All the more reason to get as much feedback from the community.
>>
>>>
>>> Also here at opaque-iri [1], is it RDFS-Plus or OWL 2 RL here?
>>
>> I wrote RDFS-Plus, because the only requirement in this test-case is 
>> to support owl:sameAs, and RDFS-Plus is the lightest regime I know 
>> that supports it. But of course, it should also hold in all variants 
>> of OWL.
>>
> I am very concerned where this is headed. RDF* should not depend in 
> any form on RDFS or OWL or anything similar.

I agree entirely that RDF* should not /depend/ on more advanced 
inference regimes. And the intention here is not at all to create this 
kind of dependency.

The meaning of the 'entailmentRegime' attribute in the test cases is as 
follow: IF your system supports that entailment regime THEN you must 
pass this test OTHERWISE you can ignore it.

The reason why these inference regime appear in the RDF* test-suite is 
that they allow to "manifest" some features of the (desired) underlying 
semantics, that should be part of the core semantics as the foundation 
for semantic extensions (such as RDFS, OWL...).

I hope this makes things clearer.

> This introduces unnecessary complications. All we need is RDF, i.e. 
> triples, graphs etc.
>
> If some here want to work on using RDF* for RDFS/OWL, they could spawn 
> off another sub-specification such as RDFS* or OWL*.
>
> Holger
>
>
>