Research On Mechanisms Of Information Propagation And Control Strategies In Social Networks

As the ideas and technologies behind Web2.0 are maturing, social media is increasingly being acquainted, comprehensible, well-known, recognized and applied. Social media, represented by Facebook, Twitter, Sina Weibo and WeChat,has changed our traditional life style, especially become a platform for sharing life, states, opinions, experience and interests. The internal structure of social media is defined as social network, which is famous for its large number of users,rich data types and rapid spread of information propagation. The massive amounts of information data are showing a trend of cross-propagation, where the speed and the potential audiences of information propagation are exponentially growing. The social network are therefore igniting both the passions from industry and academia.Firstly, whether we can observe, trace and accurately analyze the data from information propagation in social networks has become a top priority of the researches on social networks. Secondly, whether existing models of information propagation can accurately profile the rules of information propagation has become our major focus. Moreover, whether the model of information propagation can be perceived, whether the behavior of information propagation can be predicted, what features may significantly influence the predictions of information propagation and whether the predictions of information propagation can be modeled are all becoming open and challenging problems. In addition,whether the information propagation in social networks can be controlled, how to minimize the cost of information control, and how to choose the highly-influential “super” users to seriously disrupt a large portion of social networks are all worth exploring. Last but not the least, how to come up with aperfect combination of these individual methods and integrate it into the practical application system and how to develop a comprehensive analytic and control system for information propagation both need further validation.Therefore, based on the above backgrounds and relevant issues, we conduct our investigation in the following steps:1. Research on the rules of information propagation in social networks. This paper analyzes the impact of social network structure on the mechanism of information propagation, and then proposes an algorithm for generating the information propagation tree based on child dissemination recursion in social networks. Based on the observation of the real Sina Weibo datasets, seven information propagation models in social networks are induced and summarized.Furthermore, based on the analysis of the differences and similarities between the processes of information propagation and epidemic model, this paper proposes a novel rumor spreading model, called the SPNR model, by splitting the infected states with two types of infected states.2. Research on the prediction of information propagation in social networks.This paper first introduces the basic idea for perceiving information propagation and further determines the relationship between information propagation model and modularity. Moreover, this paper proposes a novel method which combines modularity with triple exponential smoothing model to perceive information propagation. In addition, this paper proposes a novel definition for event outbreaks that takes into account structural changes to the network during information propagation. This paper also investigates the main factors that contribute to event outbreaks, including structural features, temporal features,user features, and content features. This paper then constructs the model for predicting event outbreaks.3. Research on the controllability of information propagation in social networks. This paper confirms that not only cascading failures have occurred in many traditional networks, but also in social networks as well. And this paper elaborates the relationship between cascading failures and information control,and then shows that social networks with super users are vulnerable to cascading failures. Based on cascading failures, this paper proposes an algorithm of cascading failures with the crisis of trust in social networks. When a fewhighly-influential users are controlled by the above algorithm, the changes will lead the users to distrust their friends and cut off their relationships, which we refer to as the crisis of trust. As a result, cascading failures could spread across a large portion of network; sometimes, the whole network is controlled.4. Research on the information propagation analysis and control system.Based on the above three aspects of research results, this paper designs and implements the Woodpecker system. This paper further introduces the main modules of Woodpecker, including the module of data acquisition, the module of information propagation analysis, the module of information propagation prediction and the module of information propagation control. The experimental results shows that Woodpecker can effectively analyze, predict and control information propagation in social networks.This paper can better comprehend the rules of information propagation,more accurately predict information propagation model and event outbreaks, and more effectively control information propagation in social networks. These research results can be widely used in information propagation, business marketing, information recommendation and social management.