Previously we've taken a little look at how you can find & gather conversations about a brand, person or topic online. But once you have all this wonderful data how are you going to analyse it? Automated analysis of content may not be accurate and often it has a pre-defined structure, for example positive or negative sentiment. This is often too inflexible when you a particular hypothesis you want to test.

Update 8/12/09

Fail

The short of it

£5 on Olly to be voted off the X-Factor tonight.

The long of it

If Derren Brown can predict the lottery using some of wisdom of the crowds shenanigans, surely we can predict the outcome of tonight's X-Factor with some of that buzz research stuff I've been posting about. The value of buzz research is something people talk about a lot. Is all this content we're monitoring actually telling us something useful, or it it just guff. To try and figure this out I'm going to run a bit of a pseudo-experiment.