Multi-objects sentiment vector aggregation: An approach to extract latent cyber organizations

This paper aims at discovering latent cyber organizations in rich social media, through analysis sentiment dimension of individual's social attribute. From a sociology perspective, people who have the same point of views will have great probabilities and opportunities to aggression together. Monitoring these latent cyber organizations can help us understanding the tendency of social events. Naturally, individual will have his own opinions towards different objects, such as persons, events, organizations, or topics etc. We propose an effective framework to compute the individual's opinion towards certain multiple objects as his sentiment vector. Then, we detected latent cyber organizations based on the individuals' sentiment vector. To validate our approach, we illustrate the usefulness of our framework through an experiment in Sina microblog dataset, and the results demonstrate to be valuable.