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

hello ashok.

On 2012-08-07 15:53 , "Ashok Malhotra" <ashok.malhotra@oracle.com> wrote:
>>   Linked Data (upper case) seems to be
>> perceived by the majority of people to be based on RDF.
>Not so!  In Cloud computing, for example, all the data is stored in XML.
>A platform has a set of machines which is represented as a collection of
>URIs,
>Machines have many attributes, such as ports, that are, again, represented
>as collections on URIs.  I think of this as Linked Data.  Resources
>identified
>by URIs and collections of URIs.  I don't think it matters whether the
>syntax is
>XML, RDF or JSON.

you might think so and i certainly thought so as well, but i think what
matters is that we have a well-defined vocabulary that we are using. based
on http://dret.typepad.com/dretblog/2009/11/the-linked-data-police.html i
have decided to accept the fact that, not matter how much i wish it wasn't
the case, "Linked Data" has been claimed by the RDF community as being
based on RDF, and since then i have refrained from using that term (to
minimize confusion and unhappiness). i don't think we should spend
processing cycles defining the term. personally, i see it as being
RDF-based, and i would still claim that this is how 99% of "An
Introduction to Linked Data" presentations start.

>If the LDP WG ends up creating a recommendation on how to deal with
>collections
>of URIs in REST, for example, that will apply to XML, RDF and JSON

well, that brings us full circle. regardless of terminology, the question
is whether we optimize for certain metamodels. if we do, the majority of
the WG seems to favor RDF. if we don't, we should have a serious look at
existing standards (and then define patterns for using them in case you
happen to run an RDF-based back-end) instead of inventing new ones.

cheers,

dret.

Received on Tuesday, 7 August 2012 14:12:26 UTC