Identifying comprehensive personality profile from multiple online media usage

Abstract

Online media platforms have become the most popular tools for social interaction and sharing information in present world. With the increasing popularity of online media and varying needs of people using it, different media platforms have emerged. Previous research has shown some relationships between users’ personality and online media usage. Personality is shown to be useful in predicting social behavior, generating online feed, designing user interface, detecting virtual community, etc. Until now, personality is identified only from single social networking site by processing whole textual contents of online usage. This does not give a comprehensive picture of a user as different media platforms reveal different aspects of personality. In this thesis, we identify a person’s comprehensive personality profile from two types of data independently- a person’s posts in two major online media platforms (i.e., the most widely used social media platform Twitter and a major online comment posting platform Disqus) and a person’s topic related data (i.e., the person’s list of active topics and sentiment identified from online usage). Identifying personality from a person’s topic related data allows us to predict personality without processing entirety of online posts and comments which is particularly useful in cases where a person’s social media data is not directly available. In this thesis, we describe the type of data collected, our methods of analysis, our proposed novel approaches and the machine learning techniques that allow us to successfully predict personality. Our method of identifying personality outperforms existing methods which justifies the inference that single online media analysis is not enough to build a comprehensive virtual identity of a person.