The identification of personality by auto- matic analysis of conversation has many applications in natural language process- ing, from leader identification in meetings to partner matching on dating websites. We automatically train models of the main five personality dimensions, on a corpus of conversation extracts and personality ratings. Results show that the models per- form better than the baseline, and their analysis confirms previous findings link- ing language and personality, while re- vealing many new linguistic and prosodic markers.

Click Here to View the Paper In XML Format