Re: How to specify OWL version in a document

You mean to enable hilog style inference?

That is a reasonable use case though in general the philosophy is that the user picks the reasoner. In spraql we discussed quite a bit annotating queries and parts of queries with intended entailment regimes, but it was always punted to the choice of endpoint.

Since we now (thanks to Glimm et al) have some reasonable implementation techniques for the decidable fragment of hilog semantics for punned owl DL, we could potentially role it into the basic features of that profile.

I suspect that the use cases for the key bit of hilog semantics is still so few in practice that just using human oriented annotations and as hoc configuration is good enough. Having annotations that have an effect in some cases but not in others seems sorta dangerous.

On Jun 23, 2015, at 10:57, "Tara Athan" <taraathan@gmail.com<mailto:taraathan@gmail.com>> wrote:

What about the case where the ontology meets the DL syntactic restrictions but has punning? Isn't it necessary to indicate whether entailment should be based on OWL-DL or OWL Full?

Tara

On 6/23/15 12:52 AM, Bijan Parsia wrote:
On 22 Jun 2015, at 16:06, Simon Spero <<mailto:sesuncedu@gmail.com>sesuncedu@gmail.com<mailto:sesuncedu@gmail.com>> wrote:


There is no pre-defined annotation to indicate the intended OWL profile for an ontology document.  However, one can indicate actual compliance with OWL 2-DL by obeying all the DL restrictions 👮.

You can run your document through a validator as well.

In general, an annotation can become wrong for a lot of reasons. The way to be inside an intended profile is to conform to that profile’s grammar.

It is conceptually confusing to state that an ontology document is in a specific profile,  since most of the more problematic DL violations may only show up when the entire imports closure is considered - a document that imports only DL ontology documents may require OWL Full.

This is one of my favourite things :)

You can check an ontology for compliance with a given profile using the Profile validators in  OWLAPI.

Indeed.

There are existing canonical IRIs for most (all?) OWL entailment regimes defined as part of the SPARQL 1.1 recommendation series. These could be used as values for an ontology annotation.

For the reasons you stated, I would generally suggest that this would be unwise unless it’s the *result*of a checking process. But then, it’s easy enough to check, I’m not sure what value it has.

You could imagine having  a hint as to intended profile so that a lint or preflight tool can say “Hey, you are violating your intended profile!” but I doubt that’s really necessary.

Such an annotation should always be validated before making the ontology document visible.
If you are using OWLAPI , you could use such an annotation to select a profile checker. The profile checkers are not designed for incremental use, but I don't think this would be prohibitively difficult to implement.

It is possible for two documents in an imports closure, with different declared

? You mean they conform to a grammar?

sub DL profiles,  to not require  DL, as long as they don't use any clashing features.

The condition is a bit stronger: They both have to be in *one* profile. If they start out in different profiles then I don’t think they can collapse to one without already being in that one.

Note that there are documents that are in both EL and RL (or EL, RL, and QL). That is, they aren’t mutually exclusive.

Cheers,
Bijan.