The filtering approach (taken by Proxlet, for example) is pretty much the only scalable way to raise the signal-to-noise ratio of a Twitter stream. At first it seems hack-y and inelegant, but by using it in the field I've found that it is super effective.
I use a custom filtering script and a long list of regexes I've accumulated over time, expressions like this:
I use a custom filtering script and a long list of regexes I've accumulated over time, expressions like this:
\b#?NFL|NBA|NHL|World Cup|Yankees|Jets|Pats|Redskins|49ers|Steelers|Ravens|Packers|Reds|Phillies|Patriots|\ Seahawks)\b
Another filter that comes in handy is not('http://). Some Twitterers out there share good links, and that's all.