Re: Modularization [Was - Re: LDP interfaces in Java (based on Jena and JAX-RS)]

On 8/7/12 8:31 AM, Wilde, Erik wrote:
> hello kingsley.
>
> On 2012-08-07 14:18 , "Kingsley Idehen" <kidehen@openlinksw.com> wrote:
>>     A triple pattern (which isn't an RDF invention) is a powerful
>>     vehicle for data representation. Nothing about that is platform
>>     specific. The schema is fundamentally conceptual and grounded in
>>     first-order logic.
> i am amazed how on the one hand you say that RDF is not what you prefer,
> but then you jump to the conclusion that a triple-based metamodel is
> naturally the appropriate metamodel choice we should be making.
> http://dret.typepad.com/dretblog/2009/08/data-models-metamodels-cosmologies
> .html is where i tried to put all of that into context a while ago.
> briefly said: there is no "best metamodel" or "right metamodel". many
> applications happily live with tree-based or relational metamodels,
> because these are a good fit for the models of the applications. in the
> end, pretty much any model can be based on pretty much any metamodel, but
> some of these combinations work better than others, and there is no single
> best answer which one works best for a concrete problem. so what i am
> wondering about is how you arrive at the conclusion that RDF is not such a
> great metamodel, but a slight generalization of it is.
>
> cheers,
>
> dret.
>
>
>
For Linked Data the model is intrinsically an entity-attribute-value 
model enhanced with de-refrencable URIs.

Bite on my suggestion re. RDF decoupling and the REST will fall into place.

As I said, the "deceptively simple" task I see boils down to decoupling 
Linked Data from RDF.

Please walk through this sequence:

1. dereference the URI: 
http://dbpedia.org/resource/Entity–attribute–value_model
2. goto the footer of the document presented in your browser
3. note the various data representation formats.
4. execute this from your command line: curl -I 
http://dbpedia.org/page/Entity–attribute–value_model
5. also dereference; 
http://linkeddata.informatik.hu-berlin.de/uridbg/index.php?url=http://dbpedia.org/page/Entity–attribute–value_model&acceptheader=&useragentheader= 
.

The EAV/SPO based model is distinct from the data representation 
syntaxes and serialization formats. That's Linked Data. Attempting to 
redefine this as something else ultimately delivers no tangible value 
re. challenges associated with:

1. data object identity
2. data object representation
3. data object access.

The only issue (IMHO) is the conflation of implementation details (e.g. 
RDF and SPARQL) with the model and methods for data access and 
publication espoused by TimBL's original meme.

Please note, the Web has always been about Linked Data even though its 
bootstrap was primarily driven by Linked Documents (format: HTML).

-- 

Regards,

Kingsley Idehen	
Founder & CEO
OpenLink Software
Company Web: http://www.openlinksw.com
Personal Weblog: http://www.openlinksw.com/blog/~kidehen
Twitter/Identi.ca handle: @kidehen
Google+ Profile: https://plus.google.com/112399767740508618350/about
LinkedIn Profile: http://www.linkedin.com/in/kidehen