Venue: The Q Step lab (Social Sciences Library)
In this seminar we review the basic principles of text analysis in the context opinion mining of social media data, i.e. the case where the signal to noise ratio is usually low. After a brief excursus on different techniques of text analysis we present in details one specific approach called "ReadMe", due to Hopkins and King (2010). This technique has proven to be highly efficient in the context of social media analysis. Contrary to all other methods, the ReadMe approach focuses on the estimation of the aggregated distribution of the opinions rather than individual classification of texts. This allows for great accuracy in the final estimation at the cost of loosing individual classification properties, but this is not a real issues in social science research as we will show through several examples.
Some improvements over the original ReadMe algorithm will also be presented.
Daniel Hopkins and Gary King, 'A Method of Automated Nonparametric Content Analysis for Social Science', American Journal of Political Science, 54, 1 (January 2010): 229--247.
Those who are interested should contact Andrea Ruggeri (firstname.lastname@example.org)