Re: Jena implementation report plans

From: Dave Reynolds <der@hplb.hpl.hp.com>
Subject: Re: Jena implementation report plans
Date: Thu, 04 Sep 2003 15:29:49 +0100

[...]

> Let me make a preliminary comment now, to test reaction.
> 
> Several of the test cases for axioms such as the cardinality axioms implicitly 
> require comprehension axioms as well. This raises issues with systems, such as 
> Jena, which expose inference services at the RDF level.
> 
> To be concrete, consider, for example, FunctionalProperty/Manifest005. Using 
> abstract syntax for conciseness, this is currently formulated as:
> 
>   Premises005:
>    Individual(foo:object, type(owl:Thing))
>    ObjectProperty(foo:prop, Functional)
> 
>   Conclusions005:
>    Individual(foo:object, type(owl:Thing)
>               type(restriction(foo:prop, maxCardinality(1))))
> 
> An alternative formulation of this test case is:
> 
>   Premises005-mod:
>    Individual(foo:object, type(owl:Thing))
>    ObjectProperty(foo:prop, Functional)
>    Class(foo:compClass1 complete
>                restriction(foo:prop, maxCardinality(1)))
> 
>   Conclusions005-mod:
>    Individual(foo:object, type(owl:Thing) type(owl:compClass1))
> 
> This alternative formulation seems to be testing the same relationship between 
> functional properties and max cardinality restrictions but is in a form that can 
> be trivially mechanically translated into a query for an RDF API.

I don't understand this distinction.  Why cannot the first formulation also
be trivially mechanically translated into a query for an RDF API in exactly
the same way that the second can?

> Approximately 10 test cases could be usefully reformulated this way.
> 
> Possible responses to this comment include:
> 1. Modify some of test cases to this simple-conclusion style.
> 2. Augment the test cases by duplicates in this style.
> 3. Ignore it and leave the test cases as is.

Given that I don't see any significant difference between the two
formulations aside from the need for comprehension inferences in the
first I don't see any benefit here apart from a test for comprehension
inferences.  

If a completely comprehensive test suite is being developed then it might
be a good idea to have tests that demonstrate the presence or absence of
the various comprehension inferences.  I think there are many other tests
that should be generated before these, however.

> Dave

Peter F. Patel-Schneider

Received on Sunday, 7 September 2003 05:12:30 UTC