Individual Behavior from Aggregate Social Media Data
Presented at the second #FAIL2015b workshop during Internet Research 16 Conference.
Presentation slides: Bastos_individual Behavior from Aggregate Social Media Data
Aggregate databases remain relatively unexplored within social science research where researchers spend considerable time collecting data and perceive aggregate databases as offering little reward and much risk. This talk reports on the problems associated with aggregating social media data in a study spanning four years of political communication on Twitter. Although the data are publicly accessible via Twitter Streaming and REST APIs, the aggregate database is considerably different from the data made public by users because it reveals new, often unanticipated information of individual behavior. The talk closes with a discussion of potential problems in exploring aggregate social media data to study individual and group behavior.