On 11/11/2011 8:45 AM, Kingsley Idehen wrote:
> Facebook support and publish Linked Data. Of course, Linked Data != RDF
> (and its family of syntaxes and serialization formats) as sole
> implementation mechanism. (01)
Yes. Linked data was the primary motivation that started the original
WWW, but the only semantics associated with the links was in the texts
designed for human readability. Unfortunately, RDF triples just have
three links with no semantic information about those links. Furthermore
a huge amount of information on the WWW is in formats like PDF, which
also have links but no *ML language around those links. (02)
> Syntax agnostic Linked Data is the way forward. Which is why it's
> wrong to conflate Linked Data (a concept) and RDF (syntax).
> Hopefully, the W3C will come to understand this critical fact.
>
> Links:
>
> 1. http://goo.gl/y7Gq4 -- What Facebook Can Teach Us about
> Bootstrapping Linked Data at InterWeb Scales
> 2. http://goo.gl/tyqyM -- note about distributed data objects
> 3. http://goo.gl/Ez3CC -- note about data objects (03)
Thanks for the pointers. (04)
I was enthusiastic about the plans for the Semantic Web to associate
explicit semantic information with the links. I said that in notes
I sent to email lists in the late 1990s. But there were serious
flaws in the design choices adopted for the Semantic Web: (05)
1. Semantic information about a link and its content would need
to be represented in the same *ML notation as the link itself. (06)
2. But the language(s) used to process the content and to define
its syntax, semantics, and pragmatics should be based on the
"best practices" that had evolved during the previous half
century of computer science and practice. None of those tools
used the *ML languages. (07)
3. The primary convention for languages embedded in web pages was
to enclose them in <script> tags. That convention worked very
well for JavaScript and for newer languages such as PHP. (08)
4. The XML idea of stating a schema in XML notation was useful
for many purposes directly related to processing XML. But the
content processed by a knowledge representation language is
*knowledge* about some subject, which is very rarely XML. (09)
RDFa is an important correction to these design flaws. It adds
typing information about the links in a triple. It explicitly
associates the information with the document or segment of a
document in which the links are embedded. And it decouples
the linking tag from any requirement to use a *ML language
for processing the links. (010)
JSON was an obvious choice for the web companies because it is a
subset of the most widely used language for processing web pages:
JavaScript. JSON is basically LISP with curly braces, and many
kn. rep. languages have used LISP notation for over half a century. (011)
JSON also solves another major design flaw of RDF: the limitation
to triples. It's possible to represent an arbitrary SQL row, table,
or even an entire RDB in JSON notation. And there is no need to
use any reification or other transformations. (012)
And by the way, when I mentioned successful web companies that
don't use RDF, I forgot to mention Amazon.com, which uses RDBs
for their own data. They support XML for vendors to transmit
information to and from Amazon. But they don't use RDF. (013)
Unfortunately, Amazon doesn't (yet) support JSON. But various
people have provided tools to transform Amazon's XML to JSON. (014)
John (015)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J (016)
|