Now is the time to tune your tagcloud

Here at TechRepublic, we're not ashamed to admit when someone else designs something cool. Case in point: tagcloud,

an odd little online service that will accept RSS feeds, and parse

out the relevant keywords from the syndicated content. You can create

multiple "tag clouds" (like the one on our own discussions door, which

ranks user-submitted discussion tags), based around themes. I use it to

help me keep up on odd subjects, like general tech trends, the

buzzwords in the sci-fi community, and sports.

Now, like most such technologies, there's no substitute for actually

reading all the feeds you subscribe to. The tech is really bad at

filtering out signal-to-noise. Tagcloud tries to help that by letting

you exclude keywords from the tag ranks lists. For example, in my

sports cloud, the raw keyword matching will rank "coach" as the most

common word, because there's rarely a sports article that doesn't

include the term. However, that very commonality makes the term

irrelevant. So, I exclude "coach" and suddenly everything comes into


I also have clouds that use the Fark and boingboing

feeds, to help me keep up with current events. Tuning these is a little

trickier, since Fark has labels for all of its posts--Amusing, Spiffy,

Stupid, Obvious, etc.--that flood the keyword parser. Pretty much every

feed has some "always there" component that skews the results, and

tuning a cloud with multiple feeds can get tricky. How do you know when

you've got the signal-to-noise ratio right? You don't, unless you know

what the signal should sound like.

Sadly, for the moment, we do. The London bombings are dominating online

discourse, so when "London" reaches prominent keyword status in your

tagcloud, odds are you've tuned it right. Small comfort, considering,

but at least there is a small silver lining to be found.