Search Mailing List Archives
[protege-owl] properties on classes?
mccallum.anthony at gmail.com
Thu Jan 21 16:08:43 PST 2010
Does Protege 4.0 support OWL-Full? If not using OWL-Full, then the
only way to give classes data properties is using punning, correct?
Based on your explanation, annotation properties are not one really
what I'm looking for.
On 20-Jan-10, at 2:03 AM, protege-owl-request at lists.stanford.edu wrote:
> Message: 4
> Date: Tue, 19 Jan 2010 17:27:36 -0800
> From: Thomas Russ <tar at ISI.EDU>
> To: User support for the Protege-OWL editor
> <protege-owl at lists.stanford.edu>
> Subject: Re: [protege-owl] properties on classes?
> Message-ID: <09EC2187-4600-4951-B2E7-8A42337B79CD at isi.edu>
> Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed
> On Jan 19, 2010, at 5:07 PM, Anthony McCallum wrote:
>> Thanks for your response. As of now I am still working on a simple
>> classification which will not rely on reasoning. However, I would
>> like to ensure that this structure will allow for certain
>> characteristics to be attached to some of the classes that I have
>> defined. So perhaps in the future, we would like to be able to
>> reason and automatically classify individuals based on these
>> properties. Based on your answer, I understand that this could
>> potentially be possible if I use OWL-Full. This automatic
>> classification will not be attempted for quite some time (if ever)
>> so I believe that OWL-Full should be appropriate (assuming that new
>> reasoners, etc. will now be using OWL-Full). Please let me know if
>> I am completely wrong on this.
> There is some work on OWL-Full reasoners, but I'm not current with
> what is being done. In general, they are not as widespread because
> you run into computational issues with OWL-Full. That is why OWL-DL
> has the restrictions it has -- to make tractable, sound and complete
> reasoners possible.
>> On another note, is there any way to reason using annotation
>> properties, or do these have no logical significance?
> The classifiers won't do any reasoning with annotation properties,
> because they do not affect the structural definition of classes.
> Really only the property restrictions do that, so that is what is
> considered when performing classification (as well as asserted
> What sort of reasoning would you want to do with the annotation
> properties? And what sort of annotations do you wish to use? A bit
> of an example often makes it easier to come up with modeling
> Some common examples of annotations would be things like a
> lastChangedDate, author, etc. which you would want to attach to a
> particular class. The standard rdfs:comment is also like an
>> On 13-Jan-10, at 12:08 PM, protege-owl-request at lists.stanford.edu
>>> Message: 5
>>> Date: Wed, 13 Jan 2010 10:14:04 -0800
>>> From: Thomas Russ <tar at ISI.EDU>
>>> To: User support for the Protege-OWL editor
>>> <protege-owl at lists.stanford.edu>
>>> Subject: Re: [protege-owl] properties on classes?
>>> Message-ID: <B0435737-DC73-4A31-BC52-34C201E8538E at isi.edu>
>>> Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed
>>> On Jan 12, 2010, at 8:49 PM, Anthony McCallum wrote:
>>>> Is there any practical way to attach properties to classes? Or can
>>>> properties only be used in the context of individuals?
>>> Short Answer: Use an AnnotationProperty.
>>> Long Answer: It depends.
>>> If you want to stay within the OWL-DL level of expressivity, then
>>> are limited to using Annotation properties for attaching information
>>> to classes themselves. In OWL-DL, ObjectProperty, DatatypeProperty
>>> and AnnotationProperty are disjoint. And you can't attach Object or
>>> Datatype properties to classes or properties. You also cannot have
>>> classes or properties be property values, except for certain built-
>>> properties like equvialentClass, subClassOf, etc.
>>> If you are willing to use OWL-Full, then AnnotationProperty is not
>>> disjoint from ObjectProperty or DatatypeProperty. You can attach
>>> properties to classes and you can also uses classes as property
>>> values. Reasoners are not guaranteed to work on OWL-Full
>>> although in practice the reasoners will generally function while
>>> ignoring the OWL-Full constructs.
>> protege-owl mailing list
>> protege-owl at lists.stanford.edu
>> Instructions for unsubscribing: http://protege.stanford.edu/doc/
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the protege-owl