The damped oscillator model (DOM) and its application in the prediction of emotion development of online public opinions

Abstract Online public opinions refer to the attitude of the public towards certain events or topics in social media and have informative significance for social governance and the formulation of public policies. Previous studies have proved the important role played by online public opinions in expert systems from which the prediction of online public opinions has drawn increasing attention from scholars. Though the development of public opinion events is largely driven by the emotions of users, few studies have regarded the issues of emotion development forecasting, and existing emotional dynamic models are relatively simple, not real-time predicting, and with no considerations of people's self-emotion-changing mechanisms. In order to fill this gap, this paper proposes a Damped Oscillator Model (DOM). Compared to existing models, the proposed model has two main merits: first, the Mass-Spring-Damper system in the physics area is used to measure the self-decaying process of human emotion which has not been included in previous opinion dynamics models; second, a self-emotional adaptation mechanism is introduced and final states with no obviously established opinions can be reached, which is common in real online public opinion cases. Simulation experiments are conducted to discuss the impacts of critical parameters contained in the model on opinion dynamics. Two real online public opinion events were studied using the proposed model, with critical parameters extracted by the particle swarm optimization algorithm. The predictive results have outperformed several previous well-known models. With its real-time predictive capacity, this model can provide substantial auxiliary support to the expert and intelligent system for making wise decisions in advance, especially during fast developing public opinion crises.