Conversation Emotional Modeling in Social Networks

Abstract

Over the last years, the advent of social networks has changed the way of human communication giving users the ability to express their thoughts and opinions. In this paper, we present a work on analyzing human communication and interaction in Twitter. The aim is to get indicative factors about user's behavior as well as public stance and attitude towards various events around the globe. The methodology initially analyzes users' tweets and determines their emotional content based on Ekman emotional scale. Then, user's characteristics and behavior in Twitter are analyzed and their influence in the network is calculated. Based on tweets emotional content as well as user's influence, the conversation emotional graphs are developed to model and represent user's emotional interactions. Furthermore, we introduce a prediction method using machine learning techniques in order to discover the changes of topic emotional content during users' discussion.