Re: Proof: Linked Data does not require RDF

On 6/19/13 2:29 AM, エリクソン トーレ wrote:
>> -----元のメッセージ-----
>> 差出人: Kingsley Idehen [mailto:kidehen@openlinksw.com]
>> 送信日時: 2013年6月19日 10:25
>> On 6/18/13 7:32 PM, エリクソン トーレ wrote:
>>> An addendum to 1.:
>>>
>>> 1.1. However, useful* linked data will always be directly mappable to
>>> RDF.
>> Put differently, you produce more useful Linked Data via RDF. That's
>> something that easily demonstrable too!
>>
>>> I would be interested in seeing some linked data that is incompatible
>>> with RDF while still adhering to rules like using global identifiers
>>> and typed links.
>> Nobody is claiming that Linked Data is incompatible with RDF. The point
>> being made is that you can produce Linked Data, that 100% compliant with
>> TimBL's original Linked Data meme, without any knowledge or use of RDF.
>> That's all. None of that implies RDF is useless etc.. It simply means
>> that Linked Data and RDF aren't the same thing.
> My point was that even if the data producer doesn't know anything about
> RDF, when applying the meme he will produce something that follows
> the RDF abstract syntax.

RDF's abstract syntax is Subject->Predicate->Object. And what I am 
telling you is that it isn't a unique distinguishing feature since a lot 
of other folks are familiar with Entity->Attribute->Value.

The 3-tuple approach to relationship representation isn't a 
distinguishing characteristic of RDF. I produced an venn diagram [1] to 
make this a little clearer. I also produced a sample document comprised 
of structured data which isn't uniquely RDF [2].


> That is the strength of RDF and why I think
> it is an intrinsic part of Linked Data.

Please look at the venn diagram.

>   The exact semantics of RDF,
> its typed literals and blank nodes may be hard to explain, but the
> basic concept is not. Neither is it unique to RDF, as you pointed
> out elswere,
>
>>> * I wouldn't consider linked data with untyped links useful, but I
>>>     guess some people might...
>> Nobody has made any claims of that kind.
>>
>> BTW -- What is an untyped link? As far as I know there aren't any
>> untyped Links on the World Wide Web, it just so happens that the
>> semantics of the relations denoted by said links aren't necessarily
>> machine-comprehensible  (or interpretable) :-)
> For me HTML 4.1 links lacking a @rel or @rev are untyped.

But it isn't about HTML its about URIs. The Web is a woven together via 
the URIs that denote the following Relations:

1. linksTo -- as exemplified by HTML anchors you have 
[DocURI/URL]--(href)--->[DocURI/URL] (note: href is a linksTo Relation)
2. Denotes -- when a URI denotes an Entity there is an implicit Relation 
i.e., [Identifier]--(denotes)--->[thing] (on the Web URIs provide the 
denotation function).

> I suppose
> you could type them implicitly as rdfs:seeAlso or something similar.
> I also ment to include typed links where the type lacks, as you say,
> (RDF-)specified semantics.
>
> Tore Eriksson
Links:

1. http://bit.ly/16EVFVG -- Venn diagram illustrating how Identifiers 
(URIs), Structured Data (Linked Data), and RDF (Predicate Logic) are related
2. 
http://kingsley.idehen.net/DAV/home/kidehen/Public/DropBox/Public/Linked%20Data%20Resources/linked-data-rdf-test.ttl 
-- what makes this uniquely RDF?

-- 

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