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

On 8/7/12 10:30 AM, Wilde, Erik wrote:
> hello kingsley.
>
> On 2012-08-07 16:17 , "Kingsley Idehen" <kidehen@openlinksw.com> wrote:
>> 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.
> i really don' understand how you get to this conclusion. look at the IETF
> registry of media types and you'll see an amazingly wide array of all
> kinds of models and metamodels people have registered. you find trees,
> maybe jeni has even bothered to register her LMNL "overlapping tree"
> format, and all kinds of more generalized or more specialized data models.
> what brings you to the conclusion that media types are in one of these two
> simple classes you are listing? the media type world is so much more
> colorful than that.

A tree is a rooted graph. At the base you'll find a graph.

Data denotes observation.

Observation is comprised of:

1. subject
2. subject attributes
3. subject attribute values.

Thus, structured data representation is always about making the above 
discernible to humans and/or machines.

>
> i guess i'll stop wasting mailing list bandwidth for now, since you're
> going to be on vacation and nobody else seems to get engaged in this
> debate anyway. i am still failing to see, though, where those assertions
> you are making are coming from, and for my personal vocabulary management,
> i'll conclude that
>
> - there is the "Linked Data is based on RDF" perspective which is shared
> by most people, then

It isn't but I can't force you to accept this view point. I am confident 
that in due course you will realize why this is the case.

> - there's the "linked data is just data that's linked on the web"

Yes, and then you have to hone into the definition of data and how its 
represented.

Linked Data is about whole data representation via an 
Entity-Attribute-Value data model enhanced with de-referencable URIs. 
Linked Data uses de-referencable URIs to denote observation subjects, 
their attributes, and attribute values (optionally). It also uses the 
same URIs to identify web resources and then, via indirection (explicit 
or implicit),  it associates a URI with a web resource that bears the 
description of said URIs referent.

RDF isn't about what I describe above. It mandates use of URIs for 
denotation, that's it.

 From the RDF perspective re. Linked Data, just add de-referencable URIs 
to the mix and the two are connected, but not in an inextricable way.
> perspective of ashok that i also had for a while, and then
> - there's your "Linked Data is not RDF, but EAV" perspective, that is not
> something i had heard of before.

The position isn't uniquely mine. Deconstruct the definition of data, 
understand URIs, and digest TimBL's Linked Data meme with the knowledge 
that the current version is a revision of an original note that didn't 
mention RDF or SPARQL.


>
> cheers,
>
> dret.
>
>
>


-- 

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