Re: [Linking-open-data] Linked Data available

Hugh Glaser wrote:
> OK.
> Good question.
> Many thanks for the interest.
>
> Since http://budapest.rkbexplorer.com/ is simply a web page, there is no
> need to provide an RDF version.
> It would be nice to do so to give some metadata of the repository, and I
> expect we will do that in due course.
> So you are right, we should give a different code for
> "Accept: application/rdf+xml", although exactly which I discuss below.
>
> However, that does not really solve what you are asking.
> I think that somehow you expected the page to give an entrance, so an agent
> can start to explore it. As I said to Tim, we have declined to give that (at
> least for the moment), mainly because I am not sure what it would be.
> In some sense it is like trying to build a site map for a conventional site,
> or the entrance to dbpedia. Dbpedia gives some interesting queries (which I
> acknowledge we should provide), but is there an equivalent page we should be
> copying?
> Trying to provide an entrance would be easy if there was an index, but there
> isn't - these things are just knowledge bases. I see that
> http://dblp.l3s.de/d2r/all attempts something of the sort.
> On the contrary to you needing to eyeball, I have no expectation at all of
> people just browsing. If you (I assume your agent) wants to get at the RDF,
> then you resolve a URI, execute a SPARQL query, or find a URI on a search
> engine, such as Sindice (which I hope will get round to us soon).
> I would hope that people will decide to do things like adding URIs of
> themselves to their FOAF files. (We intend to keep the URIs persistent, of
> course.)
> I think this is raising interesting questions about the SW.
> I was tempted to just provide the URIs and SPARQL, but thought it might be
> helpful to put a page at the domain root.
> (If you do actually want to eyeball the data, then there is a paragraph
> about RKB Explorer on the main page.)
>
> So, sorry, I understand the possible frustration for people coming to find
> out what is there, but it may be a while before we solve the problem.
>
> So given that we don't provide an RDF page at
> http://budapest.rkbexplorer.com/ , what should we do?
> I think your suggestion of 406 may be right, but I would defer to others on
> which of the many possible responses is correct.
> In fact what it does is just the default for the web server, as we did not
> think about this question very hard.
>   
Yes there is an issue re. the index resource for an RDF Data Space. In 
our case, our site is in transition and the response below is absolutely 
wrong on our part (this was actually resolved a while back as Richard 
Cyganiak pointed this out in an earlier thread, and my response was the 
same :-) ).
> As input, I went around some other pages to find out what they do:
> curl -I -H "Accept: application/rdf+xml" http://openlinksw.com/
> HTTP/1.1 200 OK
>   

Again, wrong. It should be a 406 or a redirect to an RDF resource or 
SPARQL Protocol URL etc..

Continuing with OpenLink as an example, we have the concept of a Data 
Space that acts as the container of all the Web Resources we expose to 
the Web site. Thus, we leverage the association between a Data Space 
user and the resources created by the user en route to producing a 
personal URI for said user.

The URL re-write rules for what I describe will be in place as of the 
time you read this mail, so re-run curl to see this in action.

Also take a look at the following:

1. 
http://virtuoso.openlinksw.com/Whitepapers/html/VirtLinkedDataDeployment.html 
(HTML version of white paper)
2. 
http://virtuoso.openlinksw.com/presentations/Virtuoso_Deploying_Linked_Data/Virtuoso_Deploying_Linked_Data.html  
(Slidy presentation)


Thanks for pointing out our site and it triggered the execution of an 
item that's been wallowing on our todo :-) I also think we have a nice 
use case and best practices addendum in the making re. deployment of 
Linked Data.


Kingsley
> curl -I -H "Accept: application/rdf+xml" http://www.google.com/
> HTTP/1.1 302 Found
> Location: http://www.google.co.uk/
>
> curl -I -H "Accept: application/rdf+xml" http://www.google.co.uk/
> HTTP/1.1 200 OK
>
> curl -I -H "Accept: application/rdf+xml" http://dblp.l3s.de/d2r/all/
> HTTP/1.1 404 Sorry_class_map__not_found
>
> curl -I -H "Accept: application/rdf+xml"
> http://dblp.l3s.de/d2r/all/Publications
> HTTP/1.1 500 
> Unknown_column_enc_editor_in_where_clause_SELECT_dblp_pub_newdblp_key_FROM_d
> blp_pub_new_WHERE_enc_editor__0_LIMIT_10000_E0
>
> curl -I -H "Accept: application/rdf+xml" http://dbpedia.com/
> HTTP/1.1 302 Found
> Location: http://redir-error.gandi.net/
>
> curl -I -H "Accept: application/rdf+xml" http://redir-error.gandi.net/
> HTTP/1.1 200 OK
>
>
> On 6/11/07 21:12, "Kingsley Idehen" <kidehen@openlinksw.com> wrote:
>
>   
>> Hugh,
>>
>> I am explicitly requesting an RDF representation of the default resource
>> associated with http://budapest.rkbexplorer.com when I issue:
>>
>> curl -I -H "Accept: application/rdf+xml" http://budapest.rkbexplorer.com
>>
>> Thus, I would like to get an RDF Information Resource URI back. If such
>> doesn't exist then 406 will do. I want to get at the RDF data without
>> exercising my eyeballs :-)
>>
>> At a higher level (i.e. RDF aware User Agent) here are the results of your
>> current URIs:
>>
>> 1. 
>> http://demo.openlinksw.com/rdfbrowser/?uri=http%3A%2F%2Fbudapest.rkbexplorer.c
>> om
>> 2. 
>> http://demo.openlinksw.com/rdfbrowser/?uri=http%3A%2F%2Fnsf.rkbexplorer.com%2F
>> id%2Faward-0211606
>>
>>
>>
>> Kingsley
>>     
>
>
>
>   


-- 


Regards,

Kingsley Idehen	      Weblog: http://www.openlinksw.com/blog/~kidehen
President & CEO 
OpenLink Software     Web: http://www.openlinksw.com

Received on Wednesday, 7 November 2007 14:14:58 UTC