On Thursday 16 December 2004 17:10, Kai Kumpf wrote:
hi
here is an extract from a short talk i had on irc lately:
kakau does anyone present remember this posting
concerning a
wikipedia visual navigator? my very humble suggestion (about 2 mths
ago) was then bluntly rejected for no clearcut reasons or let's put it
the other way: did anyone of you once come across
http://www.visualthesaurus.com/online/ or touchgraphs wikibrowser
http://www.touchgraph.com/TGWB_101_SS.html ???
i've played around a bit with the touchgraph-thingy. it needs
some polish, but the code is quite readable and adding some
code to get its data from wikipedia whould be no problem.
hashar I did
I dont think we can use that on wikipedia servers, it
will generate too many queries for the DB servers
kakau and exactly *this* is *not* true. all you'd need to make that
work would be a wiki "action" that only listed either the hyperlinks or
the categories or both of one article at a time... look at
tgwikibrowser. it's fairly straigthforward. while you are in "visual"
mode, the load would be even less than for a full text query result
hashar that s still one more query made
additional queries to build
the tree
i think hashar is right: you can click quite a lot of such
tree-nodes within a second, so it would need more
queries to the database than simple browsing does, now.
i don't know about the database schema, so i cannot say
how expensive each of these queries would be.
kakau you
won't have one person do both at the same time: either
people navigate through "bubbles" or through text. this argument is
nil... 8) some people, just to insist a bit more, do in fact like
http://www.kartoo.com/ which is not so far from my suggestion, either i
just cannot imagine that the outside world is really busy and keen on
visual navigation just for laughs i'd rather think that the idea holds
some "extra value"... for the "customer"
the idea of navigating through dense graphs graphically is
quite appealing, but using it in practice i realized it's less
useable than one would think at first.
as an aside: i think i posted an idea about a
"semantic wikipedia" over
one year ago. the idea, sadly enough, was turned down too. now we see
categorization in full bloom. whom exactly should i address to make my
visual navigation proposal be heard?
8)
categorization as it is is a mess.
the question is, would real semantic information
(RDF-triples, or something like that) be better?
daniel
--
Ihr falschen Biederlinge die Ihr meiner lacht, wovon lebt Eure Politik seit
Ihr die Welt regiert? Vom Abstechen und Gemorde...
-- Till Eulenspiegel