Tags that appear together often could be closer to each other in two dimensions, and the third dimension could be popularity, for example. Naturally every user should have the ability to set the different ranking systems on different axes, including # claims on the tag, # votes, percent favorable, recent activity, # comments, etc.
Color and darkness could add 4th and 5th dimensions as needed. As always, if you can think of problems or improvements, feel free to counterclaim. I imagine that the first version would need to be an external app, but maybe we could start with gnuplot or something.
|
|
Discussion (6)
The dataset would be kinda biggish (for a web app), so you'd wanna truncate it after, say, the top hundred tags.
I don't know if gnuplot supports hypertext in data labels, and I haven't actually used it (I have Tecplot at work).
here is what it could look like in 2-D. I just threw the numbers into POV-Ray, but gnuplot will probably be easier once I get rolling with it.
that's neat, oscar. why do some people have spaces and some have underscores?
I think I switched to a semi-automated approach halfway through the process, or something. Short answer, just lazy.
Note that it would be much easier now with Many Eyes. (Any luck with that site on the Mac yet? The one time I emailed them, they were quick to respond, and helpful when they did FWIW).
I haven't touched my jyte code in forever, nor have I checked the APIs lately.
i haven't bothered contacting them because i haven't yet demonstrated to my satisfaction that i've actually got java doing anything at all, so i don't have any reason to believe it's their problem.