Steffen Lohmann wrote:
Thanks Kingsley,
I quickly added URIBurner as a dataset but cannot see the added value
w.r.t. the RelFinder - your Google-Apple example produces mainly
"seeAlso" links, which are not that helpful to discover new
relationships. Here is the link with the parameters:
http://relfinder.semanticweb.org/RelFinder.swf?obj1=R29vZ2xlfGh0dHA6Ly9kYnBlZGlhLm9yZy9yZXNvdXJjZS9Hb29nbGU=&obj2=QXBwbGV8aHR0cDovL2RicGVkaWEub3JnL3Jlc291cmNlL0FwcGxl&name=VVJJQnVybmVy&abbreviation=YnVybmVy&description=QSBzZXJ2aWNlIHRoYXQgZGVsaXZlcnMgUkRGLWJhc2VkIHN0cnVjdHVyZWQgZGVzY3JpcHRpb25zIG9mIFdlYiBhZGRyZXNzYWJsZSByZXNvdXJjZXMgKGRvY3VtZW50cyBvciByZWFsIHdvcmxkIG9iamVjdHMpIGluIGEgdmFyaWV0eSBvZiBmb3JtYXRzIHRocm91Z2ggR2VuZXJpYyBIVFRQIFVSSXMu&endpointURI=aHR0cDovL3VyaWJ1cm5lci5jb20vc3BhcnFs&isVirtuoso=dHJ1ZQ==&useProxy=dHJ1ZQ==&autocompleteURIs=aHR0cDovL3d3dy53My5vcmcvMjAwMC8wMS9yZGYtc2NoZW1hI2xhYmVs&imageURIs=aHR0cDovL3htbG5zLmNvbS9mb2FmLzAuMS9kZXBpY3Rpb24=&linkURIs=aHR0cDovL3htbG5zLmNvbS9mb2FmLzAuMS9wYWdl&maxRelationLegth=Mg==
Once again a demonstration that variety of link types is a big
challenge when automatically generating RDF data (link variety is
indeed important for the RelFinder).
Anyways - thanks for the pointer. We are generally interest in
integrating further datasets to the RelFinder's default installation
(as long as they produce valuable relationships). Further ideas and
links are welcome!
Steffen
I've just done a few things:
1. Added a gr:BusinessEntity triples for both Apple and Google
2. Sponged the DBpedia hosted descriptions of both
3. Used the FILTER feature of Relfinder to remove seeAlso links from the
mix .
When you refresh the URL above, I hope you see some aspect of what my
initial response and euphoria was about re. progressively generated and
enhances Linked Data Spaces.
More than anything else, in my world view, Linked Data is about:
1. Loose coupling of Information and Data
2. Ability to explore data relations through your own Context Lenses
across their various dimensions.
The steps above reduce the cost of "Subjectivity" and "Myopia", two
things that always benefit from Information bearing resources that don't
provide routes to their data sources.
Remember, information is "data in context", and it's inherently
subjective :-)
BTW - even if RelFinder only had seeAlso links, re. my original
comments, don't you see how the graph exposes erstwhile invisible data
or serendipitously unveils other relevant things (again very much
dependent on the context of the beholder). In my case, I found the
keynote-speakers node interesting (amongst others).
Kingsley
--
Kingsley Idehen schrieb:
Steffen,
Very cool!
Please add: http://uriburner.com/sparql to the default list of SPARQL
endpoints. The effect of doing this ups the implications of this tool
exponentially! Try it yourself e.g. Google to Apple (use the DBpedia
URIs).
The density of the graph, the response time provide quite an
experience to the user (especially a Linked Data neophyte).
Notes about URIBurner:
1. Quad Store is populated progressively with contributions by anyone
that uses the service to seek structured descriptions of an HTTP
accessible resource via ODE bookmarklets, browser extensions, or
SPARQL FROM Clause that references external Data Sources
2. As part of the graph construction process it not only performs RDF
model transformation (re. non RDF data sources); it also performs LOD
cloud lookups and joins, ditto across 30+ Web 2.0 APIs
3. Anyone with a Virtuoso installation that enables the RDF Mapper
VAD (which is how the Sponger Middleware is packaged) ends up with
their own personal or service specific variant of URIBurner.
Again, great stuff, this tool is going to simplify the message, a lot
re., Linked Data and its penchant for serendipitous discovery of
relevant things.
--
Regards,
Kingsley Idehen
President & CEO
OpenLink Software
Web: http://www.openlinksw.com
Weblog: http://www.openlinksw.com/blog/~kidehen
Twitter/Identi.ca: kidehen
