Re: relational data as a bona fide member of the SM

On Sun, Nov 6, 2011 at 12:28 PM, Enrico Franconi <franconi@inf.unibz.it>wrote:

> Describing correctly lists requires a non-first order language (you need
> fix-points, for that matter). RDF has a classical first-order semantics, so
> there is no way to correctly capture them.
>

I am not suggesting to extend RDF. I am just interested to understand why
things are this way
in RDF and what may be done better in the emerging KRs like RIF. Re fixed
points: I actually mentioned the ML data model in the beginning of this
thread :-)

--e.
>
>
> On 6 Nov 2011, at 17:03, Alexandre Riazanov wrote:
>
>
>
> On Thu, Nov 3, 2011 at 8:16 PM, Jeen Broekstra <jeen.broekstra@gmail.com>wrote:
>
>> On 04/11/11 11:22, Alexandre Riazanov wrote:
>>
>>  I don't have a problem with the OWA in general. The problem is the
>>> OWA is there even when you don't want it, specifically when you want
>>>  to be able to specify a piece of data completely and unambiguously.
>>>
>>
>> IMHO, this has little to do with the OWA. By all means, go ahead and
>> specify your data completely and unambiguously. Me adding additional (or
>> even conflicting) data about your data somewhere on the Web does not
>> suddenly invalidate _your_ data, or make it any less complete or
>> unambiguous. This is a provenance/trust-issue (whose data do you take into
>> account, and whose do you ignore?), not an OWA issue.
>>
>> The open world assumption is about allowing anyone to say anything about
>> everything, but it is not forcing you to take what everyone else says at
>> face value.
>>
>>
>>   With OWA, you cannot compute the length of a list because somebody
>>> else can redefine the list somewhere.
>>>
>>
>> 1) this is not true if you use the rdf:List construct, which specifically
>> models a closed list (indeed, it was introduced into RDF for this very
>> reason).
>>
>
> Assuming that the list is identified with a URI, someone can add different
> rdf:first and rdf:rest to it. Indeed, considering provenance may alleviate
> this problem sometimes. What if you want to trust both sources of
> information that define the list?
>
>
>
>>
>> 2) even if it were true, what's stopping you from just treating your
>> dataset as closed and computing the length anyway?
>>
>>
> If you have two definitions/descriptions of the list, which length will
> you report to the user? Both?
> Moreover, you can have two definitions in the same graph from the same
> source.
>
> Another situation is when your graph is growing incrementally and at no
> point in time you can assume it is complete. Then, if you have bags or sets
> described with some sort of membership
> predicates (like rdf:_1), to indicate when the set description is complete
> you have to use ad hoc tricks, like assigning the cardinality explicitly.
> It would be much easier with some general syntactic mechanism allowing to
> say "these are all the elements of the set". Sets, bags, lists
> are data structures. Why treat them as individuals?
>
> Cheers,
>
> Jeen
>>
>>
>
>
> --
> ======================================
> Alexandre Riazanov (Alexander Ryazanov), PhD
> Saint John, New Brunswick, Canada
> Skype: alexandre.riazanov
> http://www.freewebs.com/riazanov/
> http://www.linkedin.com/in/riazanov
> http://www.unbsj.ca/sase/csas/faculty.php
> ======================================
>
>
>


-- 
======================================
Alexandre Riazanov (Alexander Ryazanov), PhD
Saint John, New Brunswick, Canada
Skype: alexandre.riazanov
http://www.freewebs.com/riazanov/
http://www.linkedin.com/in/riazanov
http://www.unbsj.ca/sase/csas/faculty.php
======================================