Mining tweets on Twitter can be as reliable as conducting lengthy, labor-intensive public opinion polls.
A recent study from Carnegie Mellon found that sampling and analyzing Twitter tweets for data is as good as conducting an opinion poll in some cases.
According to the Twitter study, computer analysis of opinions expressed in a billion Twitter tweets during the period 2008-2009 found that measures of both consumer confidence and presidential job approval mirrored those published in public opinion polls.
Researcher Noah Smith believes the results show that analyzing the words found in streams of tweets could someday become an economical and fast way to measure public opinion on some issues.
However, he cautioned that data mining tools used to analyze public opinions on social networks are too new to know what extent they can be relied on for measuring public sentiment.
Smith noted that approximately 7 million Twitter tweets are posted every day, which could allow data mining pollsters to measure public opinion very quickly.
[ Source: India Times ]