Re: euler state of the art?

Dan Connolly wrote:
> On Mon, 2006-11-06 at 18:52 +0100, jos.deroo@agfa.com wrote:
> > 
> > The euler5 prolog version is best run as RESTfull webservice
> > (see again http://eulersharp.sourceforge.net/README)
> > as it needs the euler1 java version to do the N3->YAP 
>
> OK, I got Yapp installed, and I'm working on the Codd stuff...
> One quick question while I'm at it: what about built-ins
> that call out to the Web, like log:semantics? Should
> I expect that to work in this java/prolog/REST setup?

That should work fine (but was quite complicated to
implement as we have to call back euler1 with the
correct namespace prefixes). See also
http://esw.w3.org/topic/HCLS/ACPPTaskForce/LogicFramework
that runs fine and does keep track of already
fetched resources as kind of reasoner session caching..

> It would be a big help if Euler could do PAW
> proofs faster than cwm. Here's one of our examples;
> could you try to run this on Euler?

We have seen speeds with YAP that are 2 orders of
magnitude better than the java version..
Hm.. your PAW example actually doesn't work and
I think it has to do with incomplete implementation
of log:conjunction in the antecedent of a rule..

-- 
Jos De Roo, AGFA http://www.agfa.com/w3c/jdroo/

Received on Monday, 6 November 2006 21:00:13 UTC