Re: Against the notion of reification well-formed graph (i.e., atomicity)

On 23.01.2024 15:38, Doerthe Arndt wrote:

Hi Dörthe,

> I am also not sure whether we really need the well-formedness. Isn’t the 
> situation similar to lists? We have some syntactic sugar to represent 
> lists and there is nothing preventing the user from using the 
> list-predicates however he wants. So how does syntactic sugar for 
> reification (if that is what we want, I know, semantics is not agreed on 
> yet) differ from syntactic sugar for lists? Why should we add a 
> wellformedness condition to the spec for one of the two and not for the 
> other?

Lists are IMO a good example, and you're right that we don't prevent the 
user from using the predicates differently or, more likely, making 
mistakes in list creation and not realizing it. I say this because it 
has happened to me more than once and it is very hard to spot with the 
current tools.

With the Sugar proposal, ill-formed RDF is possible and I think it 
should remain possible for those who don't care much for RDF Star. We 
generally accept that in RDF 1.1, so I don't see why it should be any 
different with RDF Star, for better or worse.

Well-formedness becomes important IMO if you want to support RDF-Star as 
a first class citizen. Lists are again a good example here, I've yet to 
see a store that provides a performant list implementation and that's 
probably for large parts because you have to accept that they can be 
broken. That's not ideal when you're trying to optimize things.

But then the question arises as to what well-formedness entails and when 
to take care of it. The discussions we had were about bulk loading. You 
can't have indexes for things if you're not sure if the data is 
consistent (e.g. if you're loading it from a streaming source). So if we 
come up with a proposal for well-formedness, I assume it will be of 
interest to those who want to provide first-class support. And then we 
might indeed wonder if we shouldn't make it possible for things like 
lists to fix this as well. Again optionally, for those who care in their 
implementation.

And thinking of it I think syntactic sugar *is* well-formedness, as the 
Turtl parser would complain about any broken syntax and not accept it 
and/or refuse to serialize it to N-Triples or alike. So well-formedness 
means well-formedness for those serializations who do not have it 
built-in with via syntactic sugar.

regards

Adrian

-- 
Adrian Gschwend
CEO Zazuko GmbH, Biel, Switzerland

Phone +41 32 510 60 31
Email adrian.gschwend@zazuko.com

Received on Wednesday, 24 January 2024 05:31:17 UTC