Re: RDF Visualization

+cc:eric@w3.org (re perllib parser)

Hi Paul,

On Sun, 18 Aug 2002, Paul Prescod wrote:

> I thought you might be interested in my experiences as a mini usability
> test. Despite the problems I ran into, I'm happy with the end result.

Glad it turned out well in the end. My RDFViz stuff was always a rather
quick hack. To my knowledge it was the first GraphViz-based RDF
visualisation service. To my knowledge, it is currently the worst!

I suspect should shut it down and point to the more recent good stuff...


> #1:
>
> I am new to RDF visualization. So first I went to RDFViz and just hit
> the "Go!" button with the default options.
>
> I get a graphic. But it has no labels:
>
>  *
> http://snowball.ilrt.bris.ac.uk/rdf-viz/out/rdfviz-2002-08-19:41-0-29.gif

I believe this is because I lost the Windows partition (and mounted Fonts/
directory) that my GraphViz installation was pointed at.

> Then I try the SVG. My viewer complains about the namespace:
>
>  * http://www.w3.org/Graphics/SVG/svg-19990412.dtd
>
> I guess GraphViz needs to be updated to the latest SVG version??

I'm running a pretty old GraphViz binary, from before the prototype SVG
support made it into the main distribution. I believe GraphViz distro is
now up to date w.r.t. namespace URIs etc.

> There is a link that says: "External RDF Services: Send your URI to W3C
> SiRPAC server instead." I don't know what it means by "External RDF
> Services" but I'm willing to give it a try. I click the button.

The idea was to use the form as a hub that could submit data to
visualisation, query, etc etc services that were GETable or POSTable and
that did interesting things with RDF.

> I get back: "This form has been pre-loaded with data loaded from the URI
> supplied". Great!
>
> I hit "Send RDF to W3C SiRPAC Service."
>
> I get back a form that seems to have forgotten the RDF that the previous
> form had.

A service versioning problem, I think. The servlets at w3.org were updated
to use a more modern RDF parser (HP Labs' ARP), and I guess the HTTP
interface must've changed.

> Okay, fine, I go back and get the URI and paste it into the "Paste URI"
> field.
>
> This works nicely but it uses POST not GET so I don't get a nice
> GET-table URI which was my goal from the start. I want to be able to
> point people at a URI where all they have to do is hit a button or even
> better not even have to hit the button.

I *think* the servlet might be GET-happy, although the submission form isn't.

One RESTy issue here: running the visualiser creates a new resource
(ie. the downloadable image file(s) from graphviz). Currently these get
zapped periodically to avoid filling up limited disk resources. If the
service was made to expose GET, we'd need to hack the servlet to
transparently re-run graphviz if the cache'd output of the visualiser had
been deleted. And we might get religious about whether GET was appropriate
since it involves server-side changes. My hunch is that GET would be fine,
since the server-side change is an optimisation, but I'm not 100% sure.


>
> #2:
>
> So I go back to the beginning and try some other data at rdfviz.org
>
>  * http://www.oreillynet.com/meerkat/?_fl=rss10
>
> I guess it can't parse that document. (it must not use a real XML
> parser???)

From a brief look, it seems to be a problem with entities.  I'm running an
old version of EricP's Perllib RDF and XML parser.

>  *
> http://snowball.ilrt.bris.ac.uk/rdf-viz/out/rdfviz-2002-08-19:37-2-46.rtab
>
> Anyhow, once I gave up on a GET interface, I just resigned myself to
> cutting and pasting into the code text box and the SirPAC version works
> well.

Yup, ARP seems pretty complete.

Sorry you had such a bumpy ride with all this Paul. I ought to take more
care to gracefully retire old services, instead of leaving them hanging
around where they can cause annoyance.

Best place to start for RDF visualisation at W3C is
http://www.w3.org/RDF/Validator -- I hear there's a more i18n version in
the pipeline too...

thanks for your patience and for the feedback,

Dan

Received on Monday, 19 August 2002 01:30:26 UTC