[Attempto] nonstructural restrictions on Owl 2 axioms and AceWiki

[Kenneth Jones]

Thanks for the info.  The examples in the newer version of the W3C's working
draft are helpful.

The results of only two of the queries depend on the second sentence:  "What
precedes Element 4?" and "What precedes Element 7?".

Without the second sentence I get:

    What precedes Element 4?
        - Element 3

    What precedes Element 7?
        - Element 6

With the second sentence I get:

    What precedes Element 4?
        - Element 1
        - Element 2
        - Element 3

    What precedes Element 7?
        - Element 1
        - Element 2
        - Element 3
        - Element 4
        - Element 5
        - Element 6

Sorry I didn't make this clear.

But what you say about the responsibility resting with the Owl API and
Pellet instead of AceWiki makes sense.  I assume it's a just a matter of
time before this functionality is implemented.  At that time it will be
possible to lift the restriction placed on the sentences accepted by AceWiki
and rely on the more sophisticated checking done by the Owl API.  Thanks for
clarifying this.

-- Ken


> ----------------------------------------------------------------------
>
> Date: Sun, 2 Nov 2008 10:19:41 +0100
> From: "Kaarel Kaljurand" <kaljurand at gmail.com>
> Subject: Re: [Attempto] nonstructural restrictions on Owl 2 axioms and
>        AceWiki
> To: attempto at ifi.uzh.ch
> Message-ID:
>        <161f6fb70811020119pfeec018w12df52b587ce40e1 at mail.gmail.com>
> Content-Type: text/plain; charset=ISO-8859-1
>
> Hi,
>
> On Sun, Oct 26, 2008 at 4:20 PM, Kenneth Jones <kennethjone at gmail.com>
> wrote:
> >
> > In prior emails we have discussed the Nonstructural Restrictions on Owl 2
> > Axioms and their implications for AceWiki.
>
> The latest version of these restrictions (which are now called "global
> restrictions") is
>
> http://www.w3.org/TR/owl2-syntax/#Global_Restrictions_on_Axioms
>
> In this version, there are also more examples than in earlier versions.
>
>
> > You have eliminated one possible
> > violation of these restrictions by excluding the explicit definition of
> > transitive properties from reasoning and displaying red triangles
> preceding
> > the sentences that define them.
> >
> > It occurred to me that I could use the following set of sentences to make
> a
> > cardinality-restricted property transitive, without actually declaring
> the
> > property to be a transitive property:
> >
> > If X immediately precedes Y then X precedes Y.
> > If X immediately precedes Y and Y precedes Z then X precedes Z.
> > If X precedes Y then X is a sequence element.
> > If X precedes Y then Y is a sequence element.
> > Every sequence element is immediately preceded by at most 1 sequence
> > element.
> >
> > Where, of course, I have defined the noun "sequence element" and the
> verbs
> > "precedes" and "immediately precedes".
> >
> > AceWiki accepted all of these sentences and loaded them into Pellet.  I
> > created the seven proper names "Element 1" through "Element 7" and I
> linked
> > them with the "immediately precedes" verb.  The queries "What precedes
> > Element 1?", "What precedes Element 4?", "What precedes Element 7?", and
> > "What is a sequence element?" were all answered correctly.  I defined the
> > proper name "Element 2-2" and tried to enter the sentence "Element 2-2
> > immediately precedes Element 3" and the sentence was recognized as
> > inconsistent with the existing sentences and was displayed in red.  So
> all
> > my tests (queries) passed and everything looked great.
>
> As far as I can tell, the second sentence is not needed for these
> inferences to hold.
> So if Pellet just ignores the second sentence (because it is in
> violation with the global
> restrictions), then you still get the correct inferences. Only the
> inferences that require
> the second sentence to be present won't show up.
>
>
> > Of course, I realized that I must have violating the Nonstructural
> > Restrictions even though I hadn't created a sentence that was translated
> > into an Owl statement that contained TransitiveObjectProperty(precedes).
>  So
> > I'm worried that it might be possible for a user to unwittingly violate
> the
> > Nonstructural Restrictions, think everything is fine, and have Pellet go
> > haywire at a later time when just the right combination of axioms has
> been
> > defined.
>
> Yes, just disabling transitive properties syntactically will not work
> in general.
> It would provide a better solution if OWL API could be queried to find out
> if the ontology violates these restrictions. It wouldn't make sense to
> implement this checking in AceWiki as OWL reasoners and OWL APIs will have
> to
> provide it anyway.
>
> --
> kaarel
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.ifi.uzh.ch/pipermail/attempto/attachments/20081102/31417cba/attachment.htm