Re: LDOM algorithm (very early draft)

On 2/6/15 11:19 PM, Peter F. Patel-Schneider wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> My question concerned how LDOM could handle recursive shapes.  I see in the
> document:
>
>    Some other things that SPARQL engines need to implement:
>      Ability to execute ldom:Functions (recursive SPARQL queries), details to
> be written up
>      Some built-in functions, esp ldom:violatesConstraints, which maps to
> checkNodeAgainstConstraint here.
>
> It thus appears that LDOM is not based on SPARQL but is based instead on an
> extended SPARQL.

Yes, if a SPARQL engine claims LDOM support, then it needs to implement 
user-defined functions as well as a few built-in functions. The function 
ldom:violatesConstraints is basically the entry point into the 
validation. If a database does not support these extensions, then it is 
still possible to use LDOM via higher-level APIs such as Jena and 
Sesame. Maybe we could define an entry function such as ldom:support() 
that can be asked against a SPARQL engine to see if it supports LDOM or not.

>    The ability to handle recursive SPARQL queries may turn
> out to be a significant change to SPARQL engines.  Further, there are not
> sufficient details here to determine whether the required augmentations to
> SPARQL engines can be used to implement recursive shapes.

I thought we had discussed this already, but it will become clearer once 
we have written up the spec.
>
> The document also contains wording indicating that it is not using RDFS
> vocabulary in the way that this vocabulary is meant to be used.  For
> example, subclass determination is not done in the way specified by RDFS.

Could you clarify this? I believe the interpretation of rdfs:subClassOf 
in LDOM does not contradict with RDFS.

Thanks,
Holger

Received on Saturday, 7 February 2015 00:39:59 UTC