W3C home > Mailing lists > Public > public-owl-dev@w3.org > April to June 2005

Re: OWL Inference Engines

From: Dave Reynolds <der@hplb.hpl.hp.com>
Date: Wed, 25 May 2005 09:12:24 +0100
Message-ID: <42943368.4020203@hplb.hpl.hp.com>
To: Somaya Aboulwafa <somaya_ahmad@yahoo.com>
CC: public-owl-dev@w3.org

Somaya Aboulwafa wrote:

> I want to use OWL for representing security polices, more specifically 
> authorization and access control polices. So I have searching for OWL 
> inference engine that can reason over those polices. And I have found a 
> lot of owl reasoners:
> 
>    1. Java Theorem Prover (JTP)
>    2. Pellet
>    3. Fact
>    4. F-OWL
>    5. Jena 2
>    6. RACER
>    7. SweetRules
> 
> Have you any idea which is the most appropriate one for me??? anyone 
> have any recommendations????

A key question for you to consider is whether your reasoning falls within 
OWL/DL, requires OWL/full or you may also want to do some reasoning outside 
of OWL.

If you are specifically using OWL/DL and require guarantees of complete 
reasoning then use a DL reasoner - one of Pellet, Fact, or Racer. They have 
some differences, for example, Racer has more support for ABox reasoning 
than Fact which may be relevant for you.

If you want to express some of your problem outside of OWL then rule 
languages can be quite well suited to access control policies and so 
SweetRules, the Jena2 rules engine and the XSB/Flora machinery behind F-OWL 
may be appropriate.

If you may need something closer to full first order theorem proving then 
JTP is the closest on that list.

By the way, you can access Racer, Pellet and Fact through Jena2 (as well as 
using the builtin rule-based reasoner). So one approach might be to build a 
test example and then compare the performance and coverage of the reasoners 
empirically.

Dave
Received on Wednesday, 25 May 2005 08:12:39 GMT

This archive was generated by hypermail 2.3.1 : Wednesday, 27 March 2013 09:32:53 GMT