Dan Brickley wrote:
(cc: list trimmed to LOD list.)
On Thu, Jul 1, 2010 at 7:05 PM, Kingsley Idehen <kide...@openlinksw.com> wrote:
Cut long story short.
[-cut-]
We have an EAV graph model, URIs, triples and a variety of data
representation mechanisms. N3 is one of those, and its basically the
foundation that bootstrapped the House of HTTP based Linked Data.
I have trouble believing that last point, so hopefully I am
misunderstanding your point.
I am basically saying: N3 representation of graphs lead to the Linked
Data explosion.
Linked data in the public Web was bootstrapped using standard RDF,
serialized primarily in RDF/XML, and initially deployed mostly by
virtue of people enthusiastically publishing 'FOAF files' in the
(RDF)Web. These files, for better or worse, were overwhelmingly in
RDF/XML.
Not in my experience.
**critical correction: I should have stated N-Triples instead of N3 re.
base representation format at the foundation re. DBpedia **
The sequence went something like this.
TimBL Design Issues Note. and SPARQL emergence. Before that, RDF was
simply in the dark ages.
DBpedia project (which produced and still produces N-Triples dumps).
Cool URIs and Linked Data Deployment/Pubishin guides that added initial
incorporation of HTML descriptor pages into the mix while relegating
RDF/XM to a negotiable representation option.
Basically, without DBpedia there wouldn't be today's burgeoning Web of
Linked Data (what the world has come to sorta understand and started
using across many frontiers).
IMHO. As I see it, RDF/XML is a course/blessing. Without RDF/XML
sponging (so called "rdfization") wouldn't have been possible on the
scale we've achieve, many transformations would become more painful
etc.. (Blessing side). On the other hand putting RDF/XML in front of
people esp., those outside the core semweb community that assume RDF/XML
== RDF (broader framework comprised of markup and data model) when
talking about Graph Models is an unfortunate curse.
When TimBL wrote http://www.w3.org/DesignIssues/LinkedData.html in
2006 he used what is retrospectively known as Notation 2, not its
successor Notation 3.
"Notation2"[*] was an unstriped XML syntax ( see original in
http://web.archive.org/web/20061115043657/http://www.w3.org/DesignIssues/LinkedData.html
). That DesignIssues note was largely a response to the FOAF
deployment.
"This linking system was very successful, forming a growing social
network, and dominating, in 2006, the linked data available on the
web."
The LinkedData design note argued that (post RDFCore cleanup and
http-range discussions) we could now use URIs for non-Web things, and
that this would be easier than dealing with bNode-heavy data. Much of
the subsequent successes come from following that advice. Perhaps N3
played an educational role in showing that RDF had other
representations; but by then, SPARQL, NTriples etc were also around.
As was RDFa, http://xtech06.usefulinc.com/schedule/paper/58 ...
Education leads to boostrap. N3 and Turtle are very good in this regard.
Seeing the Triple is the key to comprehending what this whole thing is
about. I am sure my first few comments on SWEO echoed this fundamental
sentiment a few years back.
RDF/XML has always made the triple difficult to discern via human eyes
(esp. the RDF newbie variety).
Making HTML based presentations (a Chris Bizer & Richard Cyganiak
obsession at the time) of RDF based resource descriptions, as
exemplified by DBpedia is how the Web of Linked Data reached its bootstrap.
RDF/XML relegation to machine/program usage (e.g. the stuff we did/do
with our sponger cartridges) was crucial.
I have a hard time seeing N3 as the foundation that bootstrapped
things. Most of the substantial linked RDF in Web by 2006 was written
in RDF/XML, and by then the substantive issues around linking,
reference, aggregation, identification and linking etc were pretty
well understood. I don't dislike N3; it was a good technology testbed
and gave us the foundation for SPARQL's syntax, and for the Turtle
subset. But it's role outside our immediate community has been pretty
limited in my experience.
To be more precise, my view is that the Linked Data bootstrap occurred
modulo RDF/XML at the front door :-)
Ironically, HTML (those green DBpedia pages) played the most important
role of all. It allowed people to understand Linked Data via
conventional Web usage patterns i.e. put a link in the address bar and
then have something presented to you that made sense.
RDFa impact albeit very significant is a very recent occurrence in the
grand scheme of things re. Linked Data bootstrap (left most segment of
the adoption curve). Of course, RDFa is certainly vital to crossing
current and future adoption chasms as we continue to move from left to
right along the tech adoption curve.
cheers,
Dan
[*] http://www.w3.org/DesignIssues/Syntax.html
--
Regards,
Kingsley Idehen
President & CEO
OpenLink Software
Web: http://www.openlinksw.com
Weblog: http://www.openlinksw.com/blog/~kidehen
Twitter/Identi.ca: kidehen
