The proposal aims at studying authorship flows in information diffusion in social media. The proposal takes an individual level approach to understanding diffusion and authorship flow as compared to existing macro level approaches. The proposed approach is to study the impact of factors such as the type of media format, the type of topics diffused and the valence of the information in information diffusion and authorship flow. An initial taxonomy will be used to classify the type of media format that may evolve with the study. Three models will be combined in the research. The dataset to be used large-scale blogs and tweets. If successful, the project can help to establish a theoretical foundation for understanding the diffusion and authority in digital media.
The main goal of our research was to understand exactly how information diffuses through social media. By information diffusion, we primarily concentrated on three aspects: (1) how do people receive information?, (2) what do they do with that information once they receive it?, and (3) how do they decide to pass that information on to others? Within this context of research, we have several notable findings to report: (1) Surface Friendships vs. Deeper Connections: One finding that continues to emerge is that concentrating on just the surface level friendship network is not enough. On many social media networks, individuals can be "friends" and yet not really know each other. Instead of surface friendships, to make good predictions and understand information diffusion it is really necessary to examine underlying behavioral connections, i.e., who are people really interacting with? For instance, several of our papers have found that using the retweet and mention network on Twitter is more important than the following / Friendship network. Twitter and Brand Development: We also explored the relationship between social media use and the development of new brands. We examined relating social media usage by rock bands to sales and involvement by fans, we have shown that the strategy that a band adopts in engaging with their fans on social media can dramatically alter the participation by the fans both in terms of social media usage and actual record sales. Moreover, bands at different stages of development need to employ different strategies in terms of engaging their fans. Modeling User Behavior: One interesting finding with respect to modeling user behavior is that as the local network around a focal individual grows the behavior they have to engage in changes if their goal is to maximize diffusion of their messaging. In particular, we find that given a small number of followers people should maximize content curation, picking and choosing appropriate pieces of information to send on to their followers. As their follower list grows though, people no longer want to hear what other people say, but center on the thoughts of the focal individual, so content creation becomes more important. Urgent Diffusion: We define urgent diffusion as diffusion events where outside news is entering the network of diffusion at the same rate or faster than diffusion can occur within the network. We have shown that there are different scales for urgent diffusion depending on the type of event. However, all of these events can be captured within a general model. We have created both an agent-based and mathematical model that provide a good description of the diffusion of information in four different urgent events. Predictions in Social Media: We have found that in the space of cultural products, high frequency, low-cost social media such as Twitter can serve as a good proxy for modeling the demand of products such as music. This appears to work even better than traditional long-form data in terms of modeling and understanding the demand for data. Trending Topics in Social Media: We have found that the subject matter of a topic exploration interacts with the stability properties of trending topics. In other words, it is important to understand how subjective or objective a topic is before analyzing the trending sub-topics for that topic.
