Hi,
>
> Thanks for writing this up, Chris, it's good to have some examples to
> put a face on things. I corrected some minor typos (attached) so I
> could view the files in IsaViz et al.
Thanks, I was just brainstorming and didn't parsed the files with anything.
> There are minorly substantial
> changes from 'stlye' to 'style.' [n.b.: I'd like to start using our
> Subversion repository to track documents instead of shipping through
> email, but I'll be sending a separate message on that a little later].
>
> I like how the lens vocabulary is turning out, very nice.
>
> I don't agree that the alternate property ('if dc:title is missing, use
> rdfs:label') should be left to another layer. It may be a cleaner
> separation, but I think the vocabulary should assume the data comes as-is.
>
OK. Having this option also doesn't rule out using a inference layer below.
> I like the RDFPath for expressing how to get to some place in the graph,
> it eliminates the awkwardness of trying to come up with RDF to express
> the same. I don't quite like how it's used to express the alternatives
> or combinations of properties. That information should perhaps be more
> accessible instead of being 'trapped' in a path expression?
OK. Which would bring us back to PropertyBags and PropertySequences.
I think a good next step would be to have a look at the different RDFpath
proposals and select the subset of features that we are needing.
As I see it, there are the following proposals:
TreeHugger, Demian Steer, 2003-09
ReoPath, Neil Deakin, 2003-05
N3's Path Syntax, Tim Berners-Lee, 2002-03
A fragment identifier syntax for RDF, Jonathan Borden, 2001-05
Contrasting Logic Over XPath and RDF, Eric Prud'hommeaux, 2002-02
RDFPath Proposal, Aaron Swartz, 2001-06
Quick introduction to RDFPath, Stefan Kokkelink, 2001-03
(copied from
http://infomesh.net/2003/rdfpath/)
Emmanuel, do you know about any others?
We should also decide which functions we are needing, stuff like
"_at_startsWith('
http://xmlns.com/foaf/0.1/')" and how we handle conditions:
:styleDomain "_at_vCard:TEL [ * [
_at_rdf:type=<
http://www.w3.org/2001/vcard-rdf/3.0#home>] ]"
was supposed to mean:
Apply the style, on all vCard:TEL properties, which have a resource or bNode
as value, which has the RDF type vcard#home. I'm sure that there are better
alternatives to my syntax. Any ideas?
>
> The style vocabulary is good, but I think it's missing the capacity for
> substitution ('use this image when "dc:source :wikipedia"'). Offhand,
> that looks like it would require some more RDFPath.
>
I was trying to express substitution with a combination of "Do not show the
value" but "Show something else". See:
:TypeImageSubstitution rdf:type style:DisplayStyleProperty ;
:styleDomain "rdf:type [_at_rdf:type=<
http://xmlns.com/foaf/0.1/Person> ]"
;
style:attribute style:DoNotShow ;
style:displayBeforeValue [ rdf:type style:ImagePart ;
style:imageUrl <
http://www.example.org/PersonIcon.gif> ] ;
style:group :FOAFStyles .
But I agree a more direct approach might be better ;-)
> We discussed 'dynamic presentation,' does that belong in this vocabulary
> as well (whatever its form)? Emmanuel, you had some ideas on how to
> express this using layers - do you have time to maybe write them up as
> part of the vocabulary?
>
> Chris, you also raised some questions within the schemas:
>
> > Do you like this (re: fixed data in Style Vocab's Parts)?
>
> Maybe it's time to talk about how exactly templating fits into the
> picture...
>
Yes. Definitively. You are the velocity guys. How do lenses and velocity
templates fit together?
> > Is this getting too complicated (re: Table styling)?
>
> By nature, describing tables is going to be a little complex. But your
> solution looks pretty good. You have 'the columns of the table are
> specified by [a sublens];' I think you'd want to be able to specify
> whether they were columns or rows.
>
> I do think there are useful cases for declaring what to show by way of
> specifying what not to show. At least, I don't think there's anything
> lost by including such a mechanism.
I was thinking about something like a "all other properties" placeholder the
other day, which could be used for saying in a lens "Display all other
properties after the properties which are explicitly listed." or for
defining a style for all other properties. But maybe this is also a job for
RDF path.
Chris
>
> --
> Ryan Lee ryanlee_at_w3.org
> W3C Research Engineer +1.617.253.5327
> http://simile.mit.edu/
>
>
Received on Thu Sep 23 2004 - 10:17:56 EDT