| In order to comprehend the generation, evolution, and spreading of individual emotion, researchers in many fields have proposed numbers of theories and models. With the advent of social networks, more and more researchers realize that their simplicity, real-time, efficiency, and openness properties will facilitate the research, and may also provide available digital data for in-depth investigation. This paper dedicates to investigate the dynamical evolution of individual emotion in social networks, and proposes an emotion dynamics framework for that process. Firstly, the framework displays every environment ingredients of emotion generation, including events, media, and peers, each of which has been analyzed from the perspective of definition, feature, and the relation between it and emotion. Secondly, this paper defines the environment situation, and explores its perception influencers, such as features in perception subjects, objects, and context. Then, analyze the classification and attributes of emotion, and the emotional extraction method about online behaviors. Finally, propose an individual response mechanism in social networks based on the findings of analyzing existing psychological, computational, and hybrid related model and previous analysis.In addition, this paper proposes a prediction model of collective emotion in order to comprehend the dynamical evolution process. First of all, the model makes use of 9 social attributes as clustering features. Then, analyzes the clustering results and model the change patterns of average emotion of each cluster. Last but not least, predicts emotion for each cluster according to the trained model, and obtain the collective emotion by adding all results in linear weighted method. The experiments manifest that the prediction results fit the real changing pattern of collective emotion very well. Therefore collective emotion can be well predicted by monitoring few amounts of users. |
PDF Full Text Request