Timezone: »

Lexical and Hierarchical Topic Regression
Viet-An Nguyen · Jordan Boyd-Graber · Philip Resnik

Sun Dec 08 02:00 PM -- 06:00 PM (PST) @ Harrah's Special Events Center, 2nd Floor

Inspired by a two-level theory that unifies agenda setting and ideological framing, we propose supervised hierarchical latent Dirichlet allocation (SHLDA) which jointly captures documents' multi-level topic structure and their polar response variables. Our model extends the nested Chinese restaurant process to discover a tree-structured topic hierarchy and uses both per-topic hierarchical and per-word lexical regression parameters to model the response variables. Experiments in a political domain and on sentiment analysis tasks show that SHLDA improves predictive accuracy while adding a new dimension of insight into how topics under discussion are framed.

Author Information

Viet-An Nguyen (Facebook)
Jordan Boyd-Graber (University of Maryland)
Philip Resnik (University of Maryland)

More from the Same Authors