Re: [Bigdata-developers] Fwd: blank nodes in explorer UI

[Jean-Marc V. <jea...@gm...>]

BigData started recently developing a database console application, a bit
taylored after stardog it seems.
Anyway, this is a good thing, both as a vulgarization for non semweb aware
people, and because any database needs administration tools .

Now it's up to the BigData team to offer *some* way of navigating through
these blank node that are used  in real life, even if that may be
questionable as a modeling practice.

I don't have a preference yet (by poor knowledge of the BigData vast
possibilities) for one of the mechanisms you suggest.
As Deng Xiaoping once put it:
"It doesn't matter whether a cat is white or black, as long as it catches
mice."





2014-08-24 13:47 GMT+02:00 Bryan Thompson <br...@sy...>:

> Jean-Marc,
>
> The point of blank nodes is that they are distinct within each context in
> which they are used.  How would you like us to violate that in order to
> provide hyperlinked navigation?
>
> Really, I think that this is a problem for the foaf and the rdf data
> modeling community. They used a convenient mechanism for modeling complex
> objects, but it does not allow retrieval.  For example, if you plug in the
> top level uri of a foaf profile to bigdata or a web browser you will get
> something sensible.  You can also use bounded traversal (either concise
> bounded description or the RDF GASService - see the RDF GAS wiki page) to
> crawl the data stored in the server and navigate across those blank nodes.
>  But you can not navigate across blank nodes once they reach the user
> interface since they are no longer connected to the RDF data on the server
> in any useful manner.
>
> This is why people use concise bounded description (CBD).  We support
> several variations of CBD using DESCRIBE.  I believe that the query hints
> page of the wiki has information on how to control the CBD behavior.  But
> CBD will navigate across the blank nodes to resolve the resources that they
> are attached to and include those within the returned resource description.
>
> We certainly could make the hyperlink behavior of the workbench use CBD
> (via DESCRIBE), or offer it as a checkbox option for navigation. We can
> also expose some of the query hints used to control the DESCRIBE behavior
> to a little control panel to make it easier to use from the workbench.  How
> does that sound?
>
> Thanks,
> Bryan
>
> On Aug 24, 2014, at 5:58 AM, Jean-Marc Vanel <jea...@gm...>
> wrote:
>
> 2014-08-23 23:25 GMT+02:00 Bryan Thompson <br...@sy...>:
>
>> Taking this back to the list.  The semantics of blank nodes are not
>> supposed to provide stable identifies that can be round tripped through the
>> database.  I really advise and prefer the use of Uris for this purpose.
>>  Uris are intended to be stable identifiers.
>>
>
> Well, the problem is entirely within BigData(R) and its console explorer
> application.
> The wealth of public FOAF profiles do use blank nodes, and there is no
> possibility to make people rewrite them.
> On the other hand, the BigData(R) console explorer application must do
> something sensible to show blank nodes to the user.
> It's up to the BigData team to do something for that, and then other
> applications will do the same.
>
> That being said, a possibility I see is to distinguish between 2 kinds of
> bnodes:
> - 1. the ones that come from original data like public FOAF profiles
> - 2.  the ones that were created by the database, as result of inference
> or SPARQL node creation
>
> --
> Jean-Marc Vanel
> Déductions SARL - Consulting, services, training,
> Rule-based programming, Semantic Web
> http://deductions-software.com/
> +33 (0)6 89 16 29 52
> Twitter: @jmvanel , @jmvanel_fr ; chat: irc://irc.freenode.net#eulergui
>
>


-- 
Jean-Marc Vanel
Déductions SARL - Consulting, services, training,
Rule-based programming, Semantic Web
http://deductions-software.com/
+33 (0)6 89 16 29 52
Twitter: @jmvanel , @jmvanel_fr ; chat: irc://irc.freenode.net#eulergui