On 4/29/13 9:45 AM, Michael Brunnbauer wrote:
Hello Kingsley,

I work on that project and will answer your questions.

I now, I read the about page and actually hoped you would be the person responding :-)

Comments follow, inline.

On Sun, Apr 28, 2013 at 01:53:09PM -0400, Kingsley Idehen wrote:
1. What is the URL of your SPARQL endpoint? -- I ask because the results
in your SPARQL query form don't expose the actual SPARQL service
endpoint thereby not allowing me to access data using the
SPARQL-Protocol URL patterns.
The SPARQL endpoint with triples has the URL

  http://www.imagesnippets.com/sparql/images

and is updated once per day.

Great! I can now explore your data space using  SPARQL-FED [1][2].


The "always current" version with quads has the URL

http://www.imagesnippets.com/sparql/dbpedia

and uses the image URLs as named graph URIs (the default graph is a copy of
DPpedia for technical reasons).

No problem, I am set as per comments above re., SPARQL enpdoint, SPARQL-Protocol, and SPARQL-FED.


2. Are you exposing 5-star Linked Data URIs via this service? -- I ask
because I took a random URI (plus one for an image I annotated) from
your sample SPARQL query and passed it through the Vapour Linked Data
URI verifier, without successful results [1][2].

I ask the question above because you are minting URIs that denote
entities in a domain controlled by your app/service. Thus, you do have
the ability to put content negotiation to use by virtue of your URI
pattern choices. By that I mean, your content negotiation can be
explicit  (using HTTP response metadata) or implicit (hash based HTTP
URIs give you that).

For instance, I was able to figure out that by substituting ".png" with
".html"  in the URIs generated by your service is how I get to an
(X)HTML+RDFa document that exposes the RDF metadata for the image I
uploaded and annotated [3]. I successfully passed the document through
the W3C RDFa distiller to get a Turtle based description document [4].
Let me be sure I understand you: You took the image URLs and expected something
else than an image to be served at those URLs ?

No, I took the HTTP URIs that your system uses to denote the data it creates. Basically, we generally refer to these kinds of URIs as proxy/wrapper Linked Data URIs.

  Perhaps a 303 to the (X)HTML+
RDFa document when such a content type is requested ? Wouldn't a "Link:" HTTP
header with rel=meta be better ?

Ideally, both will work. The key here is to make your Linked Data as discoverable as possible, across user agent profiles and capability etc.. :-)

Great Job so far !! This cool service will also be added to the DBpedia apps collection (if not there already).

Links:

1. http://bit.ly/187J7aF -- SPARQL-FED based Linked Data browser via SPARQL query results from your data space
2. http://bit.ly/ZL1xMk -- SPARQL-FED query definitions page.

Kingsley

Regards,

Michael Brunnbauer

1.
http://uriburner.com:8000/vapour?uri=http%3A%2F%2Fwww.imagesnippets.com%2Fimgtag%2Fimages%2Fmm%40carmapro.com%2FP1010996.JPG&defaultResponse=dontmind&userAgent=vapour.sourceforge.net
-- vapour report on a sample URI

2.
http://uriburner.com:8000/vapour?uri=http%3A%2F%2Fwww.imagesnippets.com%2Fimgtag%2Fimages%2Fkidehen%40openlinksw.com%2FLinked_Data_Semiotic_Triangle.png&defaultResponse=dontmind&userAgent=vapour.sourceforge.net

3.
http://www.imagesnippets.com/imgtag/images/[email protected]/Linked_Data_Semiotic_Triangle.html

4.
http://www.w3.org/2007/08/pyRdfa/extract?uri=http%3A%2F%2Fwww.imagesnippets.com%2Fimgtag%2Fimages%2Fkidehen%40openlinksw.com%2FLinked_Data_Semiotic_Triangle.html&format=turtle&warnings=false&parser=lax&space-preserve=true


--

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





Attachment: smime.p7s
Description: S/MIME Cryptographic Signature

Reply via email to