Modelling emotional trajectories of individuals in an online chat

Abstract

Online communications involve an emotional component that influences the behaviour of internet users. Datasets on online communication allow for modelling emotional effects on individual and collective behaviour of users to make predictions about behaviour in online chat environments. One application of such models is decision-support for online bots that interact with users in real-time for studying the role of emotions in online communication, Affect Listeners, requiring short-term predictions about current participants. We describe an agent-based simulation of individuals' emotional interaction online based on automatic annotation of the affective content of exchanged messages. In particular, we focus on using this model to derive a default model, a 'personality', for new users joining an environment based on already observed users. © 2012 Springer-Verlag.