Twitter mining
Twitter is another amazing social phenomenon. Blogs aren’t as powerful as Twitter in describing apparently useless details of our everyday life. Big companies spend gazilion dollars to grab our thoughts and opinions and use that information to target marketing campaigns. In order to achieve this goal, companies use tools like fidelity cards or they even give you gifts if you fill a form with your personal data. Then there are of course many more means of legally acquire personal information.I can easily understand how important would be for one of those big companies to have their customers comunicate them what they think or what they do in their everyday life: “I’m eating bacon” or “I’m watching CSI” or “I don’t like politics” for example. If you correlate important keywords with time you can probably understand and maybe forecast trends. But until today storing this kind of information was extremely hard. Twitter succeeded where no other company has even dare to succeed.
Twitter data are easily available from their robust set of APIs so you can start today grabbing messages from the main stream of all the Twitter community at once. The actual limit of 1 request per minute cannot be considered a problem. The next step is to correlate and analyze those sentences stored in a special database (Dr. Inmon would call this a datawarehouse).
But without going enterprise, I’m wondering if such an application already exists: statistics about the most common words in the Twitter stream in the last 24 hours, or 7 days. Of course there is and is called twitterverse. As expected iPhone is one of the most common used words there days as well as “watching” which is far more present than “listening”. But there are other interesting statistics. What about twittering in the geo-spatial dimension? Here comes twittervision (don’t miss the 3D view). What can I say? Another way to offer at no cost our private life details to the next generation marketing tools.
Reblog Details
Saturday, August 18, 2007
