CMU Researchers Analyze Twitter Sentiments

摘要

A computer analysis of people's sentiments in a billion Twitter messages during 2008 and 2009 yielded measures of consumer confidence and presidential job approval similar to those of public opinion polls, according to Carnegie Mellon University researchers.rnThe findings suggest that analyzing the text in tweets could be an inexpensive, rapid means of gauging public opinion onrnsome subjects, says Noah Smith, assistant professor of language technologies and machine learning at Carnegie Mellon. However, the tools for extracting public opinion from the text of social media are crude and social media remain in their infancy, Smith says, so the extent to which these methods could supplement or replace traditional public opinion polls is unknown.