W3C home > Mailing lists > Public > public-owl-dev@w3.org > July to September 2010

Re: Tool for downsizing OWL 2 DL to OWL 2 EL

From: Alan Rector <rector@cs.man.ac.uk>
Date: Fri, 3 Sep 2010 09:15:18 +0100
Cc: Jan Noessner <jan@informatik.uni-mannheim.de>, OwlED-list <public-owl-dev@w3.org>, Luigi Iannone <iannone@cs.manchester.ac.uk>, Ignazio Palmisano <palmisai@cs.manchester.ac.uk>
Message-Id: <5AABF365-3BBF-4083-9EE9-F872E2240A2C@cs.man.ac.uk>
To: Pavel Klinov <pklinov@cs.man.ac.uk>
Jan

I think I would say that it is not only not trivial, but that the solution depends on choices that depend on your application, so it therefore can't be completely generic solution.

What I think you have to do is:

a) Firstly to identify where you have axioms that are outside the EL framework, e.g. Universals, cardinality restrictions, negation etc. Hopefully your ontology is based, at least infomrally, on a schema or set of patterns, so this will not be too difficult.

b) Decide how best to deal with each.  For example, cardinality max 1 can often be recast as a functional property, or perhaps a short hierarchy of a property with a functional child.  This inevitably means giving up some information or changing some of the mechanisms of use, so it involves choices - which is why there is no generic solution. 

c) From this process abstract a set of patterns for transformation.

d) Once you have the set of transformation patterns,  you ought to be able to implement them using the OWL patterns language, OPPL (http://oppl2.sourceforge.net/patterns/).  If it doesn't, please contact Luigi or Ignazio copied on this email. (Alternatively, of course, you can implement the transformations in other ways, but we hope OPPL will be easiest in the medium run and allow your solution to be shared with others.)

Hope this helps.

Regards

Alan

On 2 Sep 2010, at 19:16, Pavel Klinov wrote:

> Hi Jan,
> 
> I can't suggest a particular tool, but you might be interested in
> "approximating" your ontology in a less expressive language as opposed
> to simply dropping non-conforming axioms. Note, that the resulting
> OWL-EL ontology need not be a subset of your original ontology, but it
> might be "closer" (according to some semantic or syntactic metric) to
> the original ontology than any of the subsets.
> 
> AFAIK, Jeff Pan has been quite interested in approximating OWL 2 DL
> ontologies using OWL 2 QL and OWL 2 EL (he will correct me if I'm
> wrong). See, for example, [1]. So you may contact him and see if
> there's some implementation.
> 
> HTH,
> Pavel
> 
> [1] www.abdn.ac.uk/~csc280/pub/TPR2010.pdf
> 
> On Thu, Sep 2, 2010 at 5:39 PM, Jan Noessner
> <jan@informatik.uni-mannheim.de> wrote:
>> Hello,
>> 
>> We need to downsize an OWL 2 DL ontology to the expressivity of OWL 2 EL
>> with as less loss of axioms as possible. First we simply wanted to ask a
>> reasoner to do this work for us, but then we noticed that it is not that
>> trivial to ask all the different combinations of OWL 2 el axioms.
>> 
>> Do any of you know if there is a converter somewhere which can do this
>> downgrading job?
>> 
>> Thanks for any help in advance.
>> 
>> Jan Noessner
>> (from University of Mannheim, Germany)
>> 
>> 
>> 
> 
> 
> 
> -- 
> cheers,
> --pavel
> http://www.cs.man.ac.uk/~klinovp
> 

-----------------------
Alan Rector
Professor of Medical Informatics
School of Computer Science
University of Manchester
Manchester M13 9PL, UK
TEL +44 (0) 161 275 6149/6188
FAX +44 (0) 161 275 6204
www.cs.man.ac.uk/~rector
www.co-ode.org
http://clahrc-gm.nihr.ac.uk/
Received on Friday, 3 September 2010 08:15:48 GMT

This archive was generated by hypermail 2.3.1 : Wednesday, 27 March 2013 09:32:58 GMT