Inferring Group Identities Through Social Media

Wednesday, May 1, 2013 ( 5:00 pm to 6:00 pm )

Location: Babbio 104

Inferring Group Identities Through Social Media

Dr. Winter Mason, Assistant Professor, Howe School

Our self-perceptions, behaviors, and emotional reactions to events are deeply linked to the groups that are part of our self identity. Thus, creating computational tools to automatically infer group identities is extremely useful in both basic social science research and applied areas such as marketing and recommendation systems. Prior research developed methods using individuals with highly-focused political Twitter accounts and reported high accuracy in predicting group identity; in this work we test whether the same methods work with more typical users. I will present the results from this work as well as future directions to improve classification techniques for a broader portion of Twitter users and for a wide array of group identities.