Sandro Hawke wrote:
Sure; I just disagree that a browser that essentially gives a view of
one linked data portal should be promoted as a "linked data browser".
By that definition something like http://revyu.com/ is a linked data
browser.
Long ago in what used to be called the Semantic Web world, it was
thought that collecting rdf from out there and loading it into a
single store and then writing applications over it (such as CS
AKTiveSpace) constituted a Semantic Web application.
But then some time later, but also long ago, we realised that it was
only an application using semantic web technologies, as there was no
web involved. I think we are in danger of repeating this
misconception and distraction again in the Linked Data world.
Fundamentally the data that your browser works over is a single Linked
Data site. This site may have data that has been gathered from lots of
places, and URIs that reflect those places somehow in the text, but in
the end it is a single site.
I don't think I can give your browser any URI I choose that resolves
using http to a typical LD document? If not, it is not a linked data
browser.
I used to have a mantra: "Putting the Web into Semantic Web". It now
seems I need to say "Putting the Web back into Linked Data", or even
"Putting the Web into the Web of Data".
It may be we will just have to differ on this; however I would be really
interested to know if I am alone in my view -- any comments from others?
I'm with you 100% here.
Tabulator is an example of a "real" client-side semantic web browser.
At one point I had working server-side code that was similar; like
Kingsley's machine, it had a large database of site's data it had been
loaded with, BUT if you ever used a URI it hadn't already tried, it put
that URI on the high-priority harvesting queue, and (when things worked
well) had slurped in the data before you got your response -- so it
appeared to have all the LOD data in it. SPARQL servers can do that
Sandro,
too, and I imagine some do.
To clarify re. stuff related to Virtuoso re. Linked Data.
1. OpenLink Data Explorer (ODE) -- This is a Tabulator equivalent re.
core functionality as Linked Data browsers (both re. Browser Extension
option and Server hosted options) that uses the Sponger (Zitgist also
uses the Sponger)
2. Sponger -- The URI dereferencing engine (support HTTP URIs and other
schemes such as Handle System based lsid and doi) also exposed via a
REST API
3. Faceted Browser Engine -- server hosted and exposed via REST API
4. Web Content Crawler -- server hosted with hooks into the Sponger and
Virtuoso task scheduler
5. Virtuoso -- SPARQL compliant Quad Store and host of all of the above
Sherman has currently only put item 3 to use from his Virtuoso based
service.
The interesting questions is can we have stateless SPARQL servers that
distribute the query to other SPARQL servers, and what metadata do they
need to do that well?
I guess voiD is supposed to address that; I don't
know how well it does it, etc. (I haven't had a chance to follow this
work much recently.)
Yes, VoiD graphs cover that. The thing we need to do is standardize the
auto-discovery patterns so that smart federated SPARQL is feasible :-)
Example of a VoiD graph: http://lod.openlinksw.com/void/Dataset .
Kingsley
-- Sandro
--
Regards,
Kingsley Idehen Weblog: http://www.openlinksw.com/blog/~kidehen
President & CEO
OpenLink Software Web: http://www.openlinksw.com
