Good Features in OWL 1.1
OWL 1.1 is an extension to the existing OWL language. While it’s a work-in-progress, some of the described new features can be used to address immediate system development needs.
Among all the new features in OWL 1.1, I especially like the ability to define a Resource that is both an individual and an class, and the ability of create user-defined datatypes (e.g., Datatype(over18 base(xsd:integer) minInclusive("18"^^xsd:integer)))
Other related blogs:
Harry, do you have a reference for “a Resource that is both an individual and an class”? I thought that was the main bit of OWL Full (/RDF(S)) that didn’t work with DL.
Comment by Danny — January 3, 2006 @ 6:55 pm
It’s described in Section 2.4 “Meta-modelling and annotations” of the OWL 1.1 specification.
In the same section, the “punning” treatment is mentioned. I’m not too familiar with punning. After doing some Google search, I found this paper that talks a bit about this concept.
Comment by Harry Chen — January 3, 2006 @ 9:45 pm
Danny, the issue is to make classes as instances work well with OWL DL. There are several possible strategy (see the ISWC best paper by Boris Motik.) Punning is probably the easiest though there are still details.
You can do punning on your own now. Let’s say you want to put dublic core metadata on a class and you want to have axioms modeling that class. And you want to have property inheritence. And you want to stay owl dl.
Make two files, one with the DC assertions and one with the modeling axioms. The first file will have all the classes as instances and *only* as instances. The second will have them as classes and *only* as classes. Now, let’s say you want to answer the query, “Things with dc:creator bijan”. You run your query over each file separately, and then union the results.
That’s punning. Except you do it in one file. The syntactic position of the use of a term allows you to disambiguate it.
(See Alan Rector’s latest post about some of the issues with even this simple story. Note that dublin core properties are generally properties of the *symbolic artefact*.)
Comment by Bijan Parsia — January 5, 2006 @ 12:48 am
Thanks Harry, Bijan.
Comment by Danny — January 9, 2006 @ 7:28 am
[...] Short explanation of punning [...]
Pingback by Raw » This Week’s Semantic Web — January 9, 2006 @ 8:24 am