Psychological profiling through textual analysis
John Noecker, M. Ryan, Patrick Juola · Literary and Linguistic Computing · 2013
In this article, we examine the application of computational stylometry to psychological profiling. We adapt several techniques, which have proven useful for author identification to the problem of identifying an individual author’s Myers-Briggs personality type indicator from the statistical features of the text. The Myers-Briggs type indicator assigns four binary classifications to define personality type: Extrovert–Introvert, Intuitive–Sensing, Thinking–Feeling, and Judging–Perceiving. For this study, we use the Personae corpus, which consists of 145 Dutch-language texts all pertaining to a specific topic, each labeled with the Myers-Briggs personality profile of the author (Luyckx K and Daelemans W, Personae: A Corpus for Author and Personality Prediction from Text, In Proceedings of the 6th Language Resources and Evaluation Conference. Marrakech, Morocco: International Conference on Language Resources and Evaluation. 2008). Our system builds upon earlier work by Luyckx and Daelemans (Using Syntactic Features to Predict Author Personality from Text. In Proceedings of Digital Humanities 2008, Oulu, Finland: Digital Humanities, pp. 146–9.) to provide a set of best practices for personality profiling. We propose a more sophisticated modeling technique, combined with more advanced feature selection and state-of-the-art analysis methods from author identification to achieve a significant improvement over previous systems.