Personality Recognition Based on User Generated Content

Personality recognition has lots of applications, like business recommendation systems, human resource management, author identification and psychological treatment. Traditionally, identifying one's personality traits depends on questionnaire scales. With the development of the Internet, people can express their opinions, emotions, attitudes, etc., on social networks. From the viewpoint of psychology, people's behaviors and decisions are influenced by their personality characteristics, so it's convenient for us to recognize one's personality based on his social network content. In this paper, we build a new personality recognition model based on a standard dataset, i.e., the Facebook update status content. Firstly, we extract the language features via the Linguistic Inquiry and Word Count (LIWC) dictionary tool. Then all the words in the status content are trained by Google news pre-trained word2vec embedding with 300 dimensions. Next, we construct the personality recognition model by using the Convolutional Neural Network (CNN) method with the LIWC features. We conduct five binary classifications on the most widely used personality model, i.e., the big five personality that contains five traits: extroversion, neuroticism, agreeableness, conscientiousness, and openness. Finally, our experimental results demonstrate that the method we present in this paper achieves very good performances, especially on the openness trait.