ILRT: Metadata Research: Misc Demos: Rudolf RDFViz
Contact: Dan Brickley <email@example.com>
This demo provides a Web interface to an RDF visualisation service. For this we rely on the DOT graph visualisation algorithm, and the implementation provided by the AT&T GraphViz toolkit and the W3C RDF Perl parser.
2000-03-24 Added a new example based on the original diagrams in the 1989 proposal that led to the WWW
2000-03-29 Again thanks to ongoing work for
GraphViz, we now have SVG
output This uses a pre-release of the
module and is made available for testing (feedback to GraphViz
2000-03-24 RDFViz now generates a VRML view too - see this screenshot of a generated VRML visualisation of an RDF data file running in Mozilla.
To use, simply enter the URI for remote XML/RDF data below; the visualisation server will retrieve the data, parse into an RDF graph, and attempt to generate a GIF image depicting the graph. Please note that this is an early prototype and we have not yet added error-recovery features, so if the graphing fails for any reason the system will not currently offer any useful error messages. Work is in progress to fix this.
Usage: Enter URL for remote RDF in this form...
Privacy Warning: the current service implemented here will expose parsed RDF data and derrived .dot and .gif files in a public Web directory. Only use this system for files you're happy others seeing.
Some URIs to try...
To visualise RDF graphs we first parse XML/RDF input into an
intermediate representation (the xrdf2tab tool), then use the
tab2dot tool to generate a .dot file described the labelled RDF
graph. Finally, the
dot tool is used to generate GIF,
VRML etc for display.
The visualisation service is implemented by some tiny, hacky, scripts which glue Dot/Graphviz to an RDF parser. The whole thing needs rewriting, but is nevertheless available online for the curious. Feel free to pick these scripts up and use them to build something more polished; please let the RDF Interest Group know if you make anything interesting...
The VRML output was easy since all the hard work is done by the
GraphViz tools. Simple VRML was easy:
commandline flag to
dot does this :-). The VRML is
currently a 2D layout (apparently
recompiling and hacking a bit to output 3D layout with more legible
labels). But it shows promise -- if the 3D layout works, we could
see about getting GIFs (eg. representing people, projects,
organisations) reflected into the VRML version, and create VRML
anchor points at each node to aid automatic navigation around the
graph. (seen FSV?
remember the Jurassic Park filesystem
browser? what use is a filetree browser without metadata ? ;-)
An SVG output mode is also in progress for GraphViz, so when that's finished we'll automatically benefit from it. This will allow us to take an RDF data graph, feed it to RDFViz and get a 'best effort' automatic layout of the nodes and arcs, then tweak the final presentation in SVG editors.