Re: modelling issue?

Hi Paola:

If you are interested in database-like constraints, check SPIN 
http://spinrdf.org/

Many people are working on data provenance on a Web scale, and a long as 
you have quad stores / names graphs, there is no need to solve all the 
data quality and reliability problems at once. Just ignore those graphs 
that don't excess your proprietary "trustworthiness/reliability" score.

My apologies for stating clearly that I think open-ended multi-topic 
discussions are a waste of time. In my understanding, this is a mailing 
list for discussing specific issues of the Semantic Web vision or 
disseminating novel results/tools. We should stick to that usage.

Best
Martin

Paola Di Maio wrote:
> Hay Martin!
>
> thanks for reply, with v clear examples
>   
>> I think this discussion can be kept pretty short:
>>
>>     
>
> We can (try to) keep the discussion short, but we may  not resolve the issue
> unless we are prepared to go the bottom of
> things. (however tedious that may be, grab a long drink)
>
>  And I dont think we should exclude a priori that some of the issues being
> raise if superficially addressed
> are not at the root of the problems that come up elsewhere (using the 80:20
> rule, the majority of the problems are caused by a minority of issues)
>
>
> In OWL (and RDFS), domain and range specifications work different from what
>   
>> you would expect in a database world.
>>
>>     
>
> yes, I dont have a problem with that,
>
> I would also add, that the core relational database principles apply to any
> data model (not only to  a non database world)
>  such as : referential integrity
>   
>> than a reasoner would infer that JohnLennon is a bird.
>>
>>     
>
> oopps,  and probably even more serious wrong inferences than that ..
>   
>> . But you still don't know whether the disjointness axiom or the property
>> usage is incorrect.
>>
>>     
>
> And so? what would that result in? .....
>
> I agree our systems should be able to handle uncertainty, and I am sure
> ambiguity has its merits but...
> lack of referential integrity cannot be overlooked even in the open world...
>
> I am reading this, which contains related discussions
>
> Ref: http://eprints.ecs.soton.ac.uk/6649/
>  Managing Reference: Ensuring Referential Integrity of Ontologies for the
> Semantic Web
>
> Alani, H., Dasmahapatra, S., Gibbins, N., Glaser, H., Harris, S., Kalfoglou,
> Y., O'Hara, K. and Shadbolt, N. (2002) Managing Reference: Ensuring
> Referential Integrity of Ontologies for the Semantic Web. In: *13th
> International Conference on Knowledge Engineering and Knowledge Management
> (EKAW'02)*, Sigenza, Spain.
>
>
> also I dont see how what you discuss resolves the problem of having RDF used
> to represent different constructs
>
> class:relation:class
>  but also
> class:attribute:value
>
>
>
> but I admit I do not have the time (nor the competence) to work on this
> myself, however from where I stand
> I see some potential conflict and unreliability -
>
> do you want me to be able to trust and rely on the open semantic web, or
> not?
>
>
> If I am right, a lot of  RDF data that is being triplified and populates the
> web to support all the lovely LOD bubbles,
> are indeed taken by RDFizing data contained in dbases, in fact, often by
> simplying pushing a button 'make RDF'
> The data in dbases tends to be modelled according to relational principles,
> thefore referential integrity is inherent
> and thats when SPARQL queries can work a beauty,  minus that, and all we are
> left with is 'possibilties'
>
> I sure have nothing against that  :-)
>
> But please do correct me further, I am relatively new to this
>
>
> P
>
>
>