Re: Rules WG -- draft charter

> 1. Who wrote this draft? Might be a useful addition. (Entirely
> possible I missed attribution).

No, I left it out on purpose, because it's really a complex
collaboration (sounds nicer than "mish-mash") from the W3C Team,
trying to be true to what everyone seemed to want, but not everyone
who contributed got to see the changes I made before making it public,
so I can't really put their names on it.   Or something like that.  I
took the last editing pass, so any blame goes to me.   

> 2. (Perhaps in 2.2 Expressiveness) I'd suggest explicitly disallowing
> non-monotonic extensions of RDF semantics and things like weak
> negation [1], since I find them troublesome in a web (open world)
> context. Buchingae [2] has weak negation, for example.

Indeed - it didn't occur to me/us that anyone would imagine otherwise.
The idea is monotonic with classical negation (in so far as horn
clauses have negation).

> 3. End of 2.3 Syntax. Is the OWL rules syntax not independent of OWL?
> It looks like you can clearly pluck out the rules syntax from the OWL
> syntax, but it you do say that will be expanded upon.

To be honest, I don't know how independent the syntax or semantics
are.  They are expressed in a connected form right now.

One intricacy, for instance, is that the rules proposal does not
provide for existentials scoped within the consequent (which gets you
from datalog to Horn) except in so far as OWL provides an
equivalent expressiveness.

> 4. 2.4 Standard Library. log:semantics and the document() function - I
> think that needs plenty of justification. In XML contexts I see the
> need - referencing distinct documents - but in RDF documents are
> a pretty vague concept, and anyway RDF is trivial to merge
> and that can be done prior to applying rules.

We've made some good use of log:semantics in programming projects near
cwm.  And of course most (all?) Prologs have builtins for reading
files.  But I hear that needs to be explained better -- especially why
there should be a builtins for reading RDF and Rules sources seperate
from builtins for reading arbitrary sources.

     -- sandro

Received on Tuesday, 11 November 2003 16:19:27 UTC