On Tue, Apr 05, 2005 at 01:37:40PM +0200, Mathias Schindler wrote:
He had in mind something like a graphical map of all
wikipedia-articles
and their connections via links, something like this
http://research.lumeta.com/ches/map/gallery/wired.gif or
http://research.lumeta.com/ches/map/gallery/isp-ss.gif
I have no idea how to make such picture out of wikipedia-articles but I
would welcome any kind of feedback.
I've written a small Perl-based tool (mwgraph), which can create a graph file
from the article & category links in a mediawiki database.
You can download the mwgraph tool here:
http://debianlinux.net/wikicompany/
An example graph data file can be found here:
http://debianlinux.net/wikicompany/wikicompany.dot.bz2
Mwgraph may be a bit tricky to setup, since it needs several Perl modules
(see the Base.pm file for all the dependency URL's).
Mwgraph is a side-project of the Wikicompany project (
http://wikicompany.org),
which will soon be officially announced. The Wikicompany project aims to create
a free (GNU FDL) business directory containing: company profiles, job offers,
products info, and other such things.
The mwgraph tool creates persistent perl graph objects (so you don't have to
re-create the adjacency list and other things each time) and can directly
output a Dot file. I've only been using mwgraph on my own mediawiki DB's, so I
don't know if it could scale to wikipedia's metrics.
I've licensed the tool under the GNU GPL license, and would ofcourse welcome
additions to it. One thing I'd like to add to it is a sort of
shortest-path-to-the-top-category (eg. Categories), and output this info on a
category/article page (for navigational/orientation purposes).
Regards,
Jama Poulsen
http://debianlinux.net
http://wikicompany.net