Monday, June 13, 2005

Data Mining

Via Blogthenticity but originating at Business Week's new Blogspotting blog comes this posting.

Mining blogs for marketing intelligence is all the rage. But how to do it? Howard Kaushansky of Umbria Communications dropped by this week to tell us how his Boulder, Colo., company does it. Umbria has analyzed the language and syntax of thousands of blogs, and with that has devised an expert system that they say can analyze not only the opinions in the blogs, but also chart them by age and gender. So, clients such as Electronic Arts or Sprint can learn how their product launches or marketing campaigns are being received, day by day, by, say, teenaged boys or baby-boomer women. If Umbria can provide customer case studies, we'll blog a few.

Legions of grad students provide Umbria its raw material--blogs organized by age and gender. For $20 per hour, they go through blogs and categorize them. Given this raw material, the computer figures out what sets each demographic group apart.

Sure, manual labor is in there before the computer does its thing.

And if they make a mistake or two, it is really the aggregate data that matters. Statistically it will be on some solid ground when the calculation is done over 10 million blogs.

But it still is a scary thought.

Ultimately, will they need manual labor?

What do you think?

No comments:

Post a Comment