Shiny Doom Forum

[Mogri]

In case you haven't seen these yet, I think they're quite interesting (although not quite as extensive as google/msn/yahoo).

http://mooter.com (small, but visual)
http://kartoo.com (prettier visuals)
http://clusty.com (meta-engine, still clusters, but no pictures)

I got these from an SA article; although the article isn't that good, you can find it here: http://scientificamerican.com/article.cfm?chanID=sa006&colID=1&articleID=0006304A-37F4-11E8-B7F483414B7F0000

[fate]

for first link i get

exception

javax.servlet.ServletException: Servlet execution threw an exception

root cause

java.lang.OutOfMemoryError

for the rest:

i've found the visual representation of search results to be inadequate at best overwhelming at worst unless the clustering is done via the tree menu thing as some were or something similar and the actual links themselves were not images. I've had many a bad experience with graphically logical clustering(often without readily available labels, you had to mouse over to see any info about the found object) which made searching take longer for people people with normal sight and nearly impossible for those who are colorblind

i think its just really hard to represent so much info in so little space which is what graphical representations attempt to do a lot of the time
_________________
People's whole lives do pass before their eyes before they die. The process is called living.

Most of the time its easier to get along with people before you get to know them.

[Mogri]

I think mooter was okay... it's just a spider graph with one or two levels, and then clicking on a node brings you to the standard results list. I like the clustering idea, and although kartoo has a really busy display, I think it has the potential to be useful if you don't know exactly what you're looking for. Sure you could use a plain-text tree (I think clusty did that), but with a graphic you can use shading techniques for relavence and other grouping symbologies (e.g. you have 10 clusters, with maybe 3 distinct groupings of related clusters). A graphical representation is hard to do, but clustering is something I wished google had had in the past.