Re: Section 4: LDPR/non-LDPR formal definitions

On 3/25/13 6:44 PM, Erik Wilde wrote:
> hello.
>
> On 2013-03-25 15:24 , Kingsley Idehen wrote:
>> On 3/25/13 6:15 PM, Richard Cyganiak wrote:
>>> On 25 Mar 2013, at 22:01, Kingsley Idehen <kidehen@openlinksw.com> 
>>> wrote:
>>>> RDF
>>>> based Linked Data should have always had a media type which is akin
>>>> to actually doing the dishes.
>>> I think that text/turtle is that media type, or rather, can be it.
>> It could be, but I fear we would then stall its current progress by
>> requesting incorporation of Linked Data de-reference behavior
>> expectations into its current IANA registration.
>
> to me this really is the core problem that we are dancing around in 
> many places. most often, i refer to this as hyperRDF: RDF plus 
> hypermedia semantics, established in a way that clients can traverse 
> RDF-based hypermedia applications. how this is engineered (adding to 
> RDF as a conceptual model, or defining an ontology that must be 
> supported by anybody wishing to write hyperRDF clients) is an 
> interesting question (with different side effects based on the chosen 
> path), but the fundamental question underlying this so far does not 
> have a definitive answer: how do i expose RDF-based services in a way 
> so that applications are guided through the application flow by 
> following links. LDP has to answer this general problem one way or 
> another, and preferably there would have been an answer to this in 
> place that we could simply use. there isn't, so it remains a challenge 
> and one that obviously isn't trivially easy to solve.
>
> cheers,
>
> dret.
>
>
>
We have to the bite the bullet by taking the RDF based Linked Data media 
type route. We lose nothing by being as precise as possible, using 
mechanisms (e.g. IANA registration) that are already in place. Ambiguity 
always leads to problems.

BTW -- text/html vs text/plain is a very good analogy here.

Should we get going with the media type solution, we also end up solving 
another little problem re. RDF based graph expression oriented syntax 
notation and RDF graph serialization formats. If I recall, we had a 
thread brewing last week that was all about relative URIs and RDF which 
for all intents an purposes was yet another example of the ill-effects 
of conflating RDF syntax notation (for graph expression) and RDF graph 
serialization formats (graph representation).  You can use relative URIs 
in RDF graph expression (e.g., using Turtle Notation) but you absolutely 
can't have said relative URIs in an actual materialized RDF graph :-)

-- 

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