Re: RDF UI Vocab for the Linked Data era; is Fresnel a good choice?

Hello,

On 5 janv. 08, at 22:53, alan gebert wrote:

> Hi all,
>
> I have been going over the idea of implementing Fresnel for a
> particular mobile technology but, before taking the leap, I need to
> ask big brother the golden question.
>
> Given that some water has run under the bridge, is Fresnel still a  
> good choice?
> On the LOD world... What is the state of the art on RDF UI?
>
>
> PD: Off the top of my head, one technical observation regarding  
> fresnel:
>
> 1. Inline FSL/SPARQL expressions use qnames ( foaf:name ) instead of
> extended IRIs. I assume that the prefix scope is the document that
> contains them, but it implies that, to expand them correctly, I need
> to do one of the following:
> - Make the parser aware of these expressions ( ex: XSD datatype
> fresnel:fslSelector ) so it can correctly expand the qnames at parse
> time

That's what JFresnel [1] does at Fresnel parsing time, for both the  
Jena and Sesame implementations. FSL and SPARQL expressions are  
recognized as such through their XSD type by the Fresnel parser,  
which knows about prefix bindings through the Jena or Sesame API, and  
replaces the qnames by the actual URIs (actually it is a bit more  
complicated than that since you can have wildcards, but that's the  
basic idea). I guess Simile Longwell's Fresnel engine (based on  
Sesame) does something similar (for the FSL part I am sure of it  
since it uses the same FSL engine as JFresnel). But it is true that  
you might get into trouble if your framework does not let you access  
prefix bindings in the Fresnel/RDF document containing those  
expressions.

[1] http://jfresnel.gforge.inria.fr/



> - Store the required provenance info and the prefix set so the Fresnel
> engine can make the expansion later
>
> The latter seems easier to do... but still feels a bit "odd" and I can
> see myself running into some trouble. I know some parts of my current
> framework won't respect that completely ( smushing, inference, remote
> parsing... a la Triplr ) as they treat the literal as an obscure rdf
> term and may replace "foaf" for "p1" without asking anyone for
> permission.
>
> OTOH, including the prefix list everywhere is not a clean solution  
> either....
> I propose defining a new property that travels along with the group: (
> fresnel:prefixes "foaf:<...>, foo:<..>" ).
>

That's something we could discuss on fresnel-dev [2].

[2] http://simile.mit.edu/mail.html

Emmanuel

--
Emmanuel Pietriga
INRIA Futurs - Projet In Situ    tel : +33 1 69 15 34 66
Bat 490, Université Paris-Sud    fax : +33 1 69 15 65 86
91405 ORSAY Cedex FRANCE     http://www.lri.fr/~pietriga

Received on Sunday, 6 January 2008 10:41:31 UTC