Inferring User Personality from Twitter

Abstract

Personality affects how a person behaves when interacting with people and computer systems. It can determine one’s needs and preferences in different contexts, which is especially useful for adaptive and recommender systems. Personality questionnaires are widely used to acquire information about user personality. However, filling in them can be tedious. Analyzing the user interactions can be another way of obtaining that information. Since personality has an impact on the way a person interacts with others, and social networks are widely used for this purpose, information about user interactions through social networks can give a clue about their personality. In this paper, we present a system able to obtain data about user interactions in Twitter and analyze them in order to infer user personality. The system has been used not only to infer personality but also to compare and evaluate different user models, classifiers and personality dimensions