Persona Classification of Celebrity Twitter Users

Abstract

Twitter is a microblogging service allowing users to post up to 280 characters at a time describing their thoughts. Twitter currently receives about 500 million tweets a day, in which people share their comments regarding a wide range of topics. In the process of creating social network profiles, users reveal a lot about themselves in what they share and how they say it. Through self-description, status update, and tweets, we can find a lot about the users. A user's knowledge of social sites could be remarkably improved if other information like demographic attributes and user's personal interest and the interest of other users are considered. This is truer in case of celebrity users. This chapter attempts to analyze celebrity tweets to provide relevant recommendations to the practitioners. The tweets of celebrity users are classified using two distinct approaches (1) Fixed Classification into six predefined categories and (2) Generating a category if the tweet does not belong to any defined category. The first kind of classification has been done in three different ways; by individually applying Naïve Bayes, Decision Tree, and Support Vector Machine. For generating a new category, Latent Dirichlet Allocation is used. Henceforth, this Persona Classification of Celebrity Twitter Users will help users to gain insight into their interests thereby decluttering their twitter feed and showing them relevant content on their feed. With the understanding of celebrity persona, smart recommendation systems can also be designed.