Search Mailing List Archives


Limit search to: Subject & Body Subject Author
Sort by: Reverse Sort
Limit to: All This Week Last Week This Month Last Month
Select Date Range     through    

[protege-discussion] OWL2 Classes

Timothy Redmond tredmond at stanford.edu
Fri Mar 16 17:13:20 PDT 2012


On 3/16/12 1:22 PM, Jim Tivy wrote:
>
> Hi Folks
>
> I am looking to implement a Thesaurus using the OWL2 model.  It seems 
> the recommended OWL2 modelling is to use a class for each term..  When 
> I look at class, however, it seems that it is very weak because the 
> mechanism for attaching properties to classes is weak.
>

Actually I think that the OWL 2 formalism is very expressive.  The thing 
that you have to figure out is exactly what you mean by "attaching 
properties to a class".   So if you have a class, A, a class B and a 
property p you can say

  * all individuals in the class A must have a p-property value:

        A SubClassOf p some Thing

  * all individuals in the class A must have a p-property value that is
    in the class B:

        A SubClassOf p some B

  * if some individual, i, has a p-property value then the individual i
    must be an element of the class A:

        p domain A


This is only just barely scratching the surface.  So tell us what you 
are trying to express and we can start thinking about whether OWL 2 can 
express it.

> I think I am better served to have one class called "Terms" or 
> "Concepts" whose individuals express the Thesaurus.  In that way I can 
> constrain properties and reason about these individuals more naturally.
>

It may be that this is true but it seems very unlikely to me.  My sense 
is that if you don't have a rich class structure then you are not going 
to have much to reason about with the individuals.

> I realize Class==Concept in OWL2, however I think that since Class 
> itself is not an individual that it is too weak.  What ever happened 
> to the notion of the Class Class.
>

What is the "Class Class"?  If you are thinking of meta-modeling, then 
my reaction is that I think that realistic meta-modeling is probably 
often quite difficult to get right.  But in any case, the starting point 
is to figure out what you are trying to say.  Then we can figure out 
what language capabilities you need to express your concept.

-Timothy


> Jim
>
> Jim Tivy -- CTO, Bluestream
>
> Skype: jimt.vanc
>
>
>
> _______________________________________________
> protege-discussion mailing list
> protege-discussion at lists.stanford.edu
> https://mailman.stanford.edu/mailman/listinfo/protege-discussion
>
> Instructions for unsubscribing: http://protege.stanford.edu/doc/faq.html#01a.03

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.stanford.edu/pipermail/protege-discussion/attachments/20120316/2c946d36/attachment.html>


More information about the protege-discussion mailing list