A study suggests that analyzing Twitter data can provide similar results to large-scale public-opinion polls, potentially providing a cheap alternative to telephone polling. Researchers used text-analysis software to detect the sentiments expressed in a billion Twitter messages sent over a two-year period and found their results closely tracked the findings of well-reputed pollsters like Reuters and Gallup. Researchers did note, however, that Twitter data tend vary more day to day than traditional polling data, in part because analysts are still learning how best to interpret tweets.

Full Story:

Related Summaries