Re: is linked data about RDF or EAV or just structured data?

On 8/7/12 10:02 AM, Wilde, Erik wrote:
> hello kingsley.
>
> changing the subject to make it a little easier for people to ignore this
> selectively. but i am enjoying this.
>
> On 2012-08-07 15:43 , "Kingsley Idehen" <kidehen@openlinksw.com> wrote:
>> On 8/7/12 9:15 AM, Wilde, Erik wrote:
>>> understand how you end up on that middle ground (it's not RDF, but it's
>>> EAV), and saying that it is "intrinsically" like that does not really
>>> explain where this perspective originates.
>> Maybe we should look at it from a different perspective. What is your
>> understanding of structured data? In my case it boils down to data
>> representation where the relationship between an entity, its attributes,
>> and its attribute values are clearly discernible. Syntaxes for
>> expressing the aforementioned vary.
> i am a mostly a REST guy. so as a very primitive starting point,
> structured data is whatever you define as a media type. it can be as
> simple as CSV, or more complex metamodels such as XML or RDF.

Modulo RDF re. your comments above, since it isn't a format, a media 
type still boils down to an entity-attribute-value or attribute=value 
structure i.e., 3-tuple or 2-tuple. It just documents the fact in prose 
as part of the mime type.

> to be in
> line with REST, you need to make sure that the media type has hypermedia
> controls in it, and those should be based on URIs as identifiers.

Linked Data does that in the most generic way. A document can an agent 
anything it needs to know e.g., how to update, delete, insert, read etc..
>   what
> kind of URI scheme you choose is already up to you, REST just tells you
> that interactions should be driven by links, so it needs to be a URI
> scheme that clients can use to initiate interactions. that's about all you
> can say about the constraints driven by REST.
>
> your starting point of "an entity, its attributes, its attributes values"
> is already driven by more specific constraints, and i am curious to learn
> where they originate, and what they are. or asking the other way around,
> if you have a non-EAV model (such as atom) that has a well-defined
> interaction model based on hypermedia controls, does that even qualify as
> "structured data" in your perspective?

I can make Linked Data using Atom. See: 
http://linkeddata.informatik.hu-berlin.de/uridbg/index.php?url=http%3A%2F%2Fdbpedia.org%2Fdata%2FLinked_data.atom&useragentheader=&acceptheader= 
.


>
> cheers,
>
> dret.
>
>
>
I am about to head out on a road trip, so there will be some lag in my 
responses. That said, keep them coming, this is extremely important. If 
we nail this we solve a long running puzzle while also checking off an 
easy win, sorta :-)

-- 

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