- From: Tim Berners-Lee <timbl@w3.org>
- Date: Mon, 29 Oct 2007 19:10:38 -0400
- To: Sean B.Palmer <sean@miscoranda.com>
- Cc: "Yosi Scharf" <syosi@mit.edu>, public-cwm-bugs@w3.org
On 2007-10 -29, at 15:37, Sean B. Palmer wrote:
> On 10/12/07, Yosi Scharf <syosi@mit.edu> wrote:
>
>> Right here is your problem. The meaning of ?s is that the
>> variable "s" is declared one level above.
>
> Oh dear! I had known but forgotten because it's counterintuitive to me
> that it should be easier to quantify a universal over its parent
> formula than the document's root formula.
The problem with the top level is that if you take a rule and
enclose it braces you get not what you want:
# Works only with quantification in parent formula:
{<foo.n3> !log:semantics log:includes { {?x p ?y. ?y p ?z } => {?x
p ?z} }} => { TR a OK }.
# Works only with quantification in parent formula:
@forall p.
{ p a owl:TransitiveProperty } => { {?x p ?y. ?y p ?z } => {?x p ?
z} }.
There are also plenty of times when like the p above you would like
it implicitly quantified at the top level
# Works only with quantification in root formula:
{ ?p a owl:TransitiveProperty } => { @forAll x,y,z. {x ?p y. y ?p
z } => {x ?p z} }.
Which is most intuitive?
> It would be interesting to run a survey of all N3 rules files to see
> what the actual deployed usage patterns are: are universals in
> sub-subformulae more likely to be quantified over their parent
> formulae, or over the document's root formula? My guess is the latter.
>
Well, I don't know what else other than cwm handles nested formulae and
also variables. Anything cwm processes must use the n3 definition of ?
x which cwm
uses.
> If it's strongly the latter, perhaps it's worth considering changing
> this. I feel it's probably better to leave this bug open until such a
> survey is done. What do you reckon?
>
>> I'm not sure how cwm interprets when variables are declared
>> inconsistently like that. Weirness resulted.
>
If ?x is used at more than one level, an error should result.
> Yeah. I'd like to know how a log:/--think processor *ought* to handle
> strange sidecases like that, but it's academic really—some fun to take
> place should N3 ever be formalised. It may be regarded as not part of
> this bug report.
>
> --
> Sean B. Palmer, http://inamidst.com/sbp/
Received on Tuesday, 30 October 2007 18:41:56 UTC