- From: Dan Connolly <connolly@w3.org>
- Date: Thu, 14 Dec 2006 02:09:38 -0500
- To: jos.deroo@agfa.com
- Cc: public-cwm-talk@w3.org
On Dec 13, 2006, at 8:28 PM, jos.deroo@agfa.com wrote:
> I'm finding an issue with euler n3->yap interpretation
> for the http://www.w3.org/2002/12/cal/calAx.n3 rule
>
> { ?P a :DateProperty.
> ?E ?P ?WHEN.
> # we skip the dtlit step of splitting ?WHEN apart
> (?WHEN "(\\d\\d\\d\\d)-" ) str:search (?YYYY).
> (?YYYY 0) math:sum ?YEARNUM. # sum coerces strings to int
> } => { [ is k:YearFn of ?YEARNUM ] }.
>
> it actually gives a wrong intermediate N3 translation
>
> {?P a p0:DateProperty. ?E ?P ?WHEN. (?WHEN "(\\d\\d\\d\\d)-")
> str:search
> (?YYYY). (?YYYY 0) math:sum ?YEARNUM} =>
> {?YEARNUM!k:YearFn }.
That does look like a different way to say the same thing. Does
?YEARNUM!k:YearFn not result in a new bnode/skolem constant?
> hm..
> and I still have to implement str:search ..
Actually, I just took str:search out of calAx.n3 in favor of time:month
and such.
> otoh I think we achieved a stable outcome of e:findall and a correct
> e:bayesian
> in namspace http://eulersharp.sourceforge.net/2003/03swap/log-rules.n3
> resulting in over 10 billion inferences today for tests such as
> http://eulersharp.sourceforge.net/2006/02swap/etc5.ref and other
wow... I hope to take a look at that soonish.
--
Dan Connolly, W3C http://www.w3.org/People/Connolly/
Received on Thursday, 14 December 2006 07:09:49 UTC