RE: a question about reasoner

On September 4, Peter Crowther writes:
> 
> > From: Zhu Bin [mailto:zhubin@cai.pku.edu.cn] 
> >   Rule-based reasoner and tableaux-based reasoner, which is the better
> > for the reasoning in ontology?
> 
> They can do somewhat different things; you may even want to combine
> them.  If you are looking at building a reasoner for OWL-DL, a
> tableau-based reasoner would probably give the best combination of
> functionality and performance.  If you are restricted to a rule-based
> reasoner, OWL-Lite is designed to provide a good framework that is
> (supposedly) implementable using rules.

I would be interested to know the basis on which you make this last
statement - pointers to relevant literature would be particularly
welcome.


Ian

> 
> Take a look at FaCT and Racer (tableaux-based reasoners), Euler
> (rule-based reasoner) and TRIPLE (rule-based that calls out to
> tableaux-based where necessary).  Of these, I think TRIPLE is
> architecturally the most interesting.  Apologies to the folks whose work
> I have missed out.  There are more links in the Reasoners section of
> [1].
> 
> > Should we divided reasoning and query
> > into two part in ontology?
> 
> I'll leave that to others :-).
> 
> 		- Peter
> 
> [1] http://www.w3c.org/2001/sw/WebOnt/impls
> --
> Peter Crowther, Director, Melandra Limited