Re: well-formed Shape Expressions Schemas

Le 17/11/2015 15:03, Peter F. Patel-Schneider a écrit :
> It appears to me that the definition of well-defined Shape Expression schemas
> in http://arxiv.org/abs/1510.05555 is unduly restrictive.

You are right; the restriction is stronger than what is needed for 
well-definedness alone.

This restriction also guarantees that checking that a node does not 
satisfy a shape is "easy", as we only need to explore a limited 
neighbourhood of the node. That's a second reason why the restriction is 
severe.

It's on my stack to imagine a less restrictive restriction that still 
gives some guarantees on complexity of validation.

Iovka

>
> It rules out, for example
>
> <A> { ex:p @<A> }
>
> <B> { ex:q !@<A> }
>
> even though this schema is well behaved, i.e, it has a unique maximal model.
>
> So, why not allow any set of shape definitions that has stratified negation?
> Stratified negation is a well-studied notion and evaluation techniques for
> stratified negation are well known and should be easily adaptable to Shape
> Expression schemas.
>
>
>
> peter
>
>
> PS:  I think that you meant to say "acyclic directed graph" in Definition 1.


-- 
Iovka Boneva
Associate professor (MdC) Université de Lille
http://www.cristal.univ-lille.fr/~boneva/
+33 6 95 75 70 25

Received on Thursday, 3 December 2015 15:45:16 UTC