Fabio Celli , post-doctoral researcher at University of Trento, Trento, Italy
Abstract: Personality Recognition consists in the automatic prediction of people's personality traits from various sources, including text and multimedia, such as pictures, audio and video. These predictions can be exploited for a wide variety of tasks, research fields and real world applications, like recommender systems, user profiling, marketing, HR and security. Personality recognition has grown considerably in the last decade as a research field, and has been boosted in recent years by the availability of large databases of self-assessed data. Gold standard labels for an objective evaluation of personality types can be obtained by means of the Big5 personality tests, which are well-known and widely accepted in psychology and other research fields. The Big5 factor model defines personality along 5 bipolar scales: extraversion (sociable vs. shy); emotional stability (secure vs. neurotic); agreeableness (friendly vs. ugly); conscientiousness (organized vs. careless) and openness to experience (insightful vs. unimaginative).
From a computational perspective there are three main approaches to the recognition of personality from text: top-down, bottom-up and mixed, and each one has advantages and disadvantages when applied to specific tasks. In this talk we will present an overview of the three approaches to the computational recognition of personality, we will see the results of evaluation campaigns and applications in social media and security, in order to provide insights on what works and what doesn't, under which conditions. Bibliography: