Micro-blog Rumor Diffusion Model And Empirical Research Based On Complex Networks

Based on the complex network, this paper study the rule of Micro-blog rumor diffusion with methods of model and empirical analysis. With the widespread use of digital technology and Internet technology, Internet as a new kind of media for information publishing and dissemination is having a greater influence on the spread of rumors than ever before. Especially on the social media, such as Micro-blog, information has fission dissemination characteristics and rumors can spread exponentially. Due to the diversity, interactivity and individuation characteristics of information in cyberspaces, it hard to reveal the rules of the diffusion of Internet rumors through content analysis and text analysis. Complex network, a new tool for quantitative description of complex system, provides a new theoretical perspective and method for the research of rumor diffusion which has network structural features.In this study, individuals and their relationship on Micro-blog are abstracted as nodes and edges. And the forwarding chain of a single micro-blog is regarded as research object and used to build a network of rumor diffusion based on forwarding relationship. Firstly,100 Micro-blog judged as false information by Micro-blog Community Management Center are collected and classified into 10 categories based on their content. By computing the statistical variables of network topological features of these Micro-blog samples, including degree, average path length, clustering coefficient, modularity, etc, we find that Micro-blog rumor diffusion networks have small world and scale-free characteristics. Using correlation analysis and cluster analysis, we classify those network variables into three types of indexes:the network size,the nodes’distance and the density, and divide Micro-blog samples into 6 groups by different network structures according to these indexes.Secondly, borrowing ideas from the classical epidemic model—SIR model, and taking the information uncertainty and individuals’ attitude into consideration, we build a transmission dynamics model called SIpInR model in which individuals in rumor network can exist four discrete states, namely, susceptible, positively infected, negatively infected and resistant. Netlogo, a multi-agent simulation platform, is used to conduct simulation experiments to study effects on rumor diffusion through changing the parameters of infection rates, transfer rates and immunization rates. Then we verify the validity of the model through comparing the results of simulation experiments with real Micro-blog data.Finally, through counting statistics of real Micro-blog data, this paper measures the importance of nodes in Micro-blog network with the degree centrality, the betweenness centrality and the closeness centrality. And then it also explores the correlation between different centrality. On the same time, introducing immunization mechanism into the model above, we test the effects of different immunization strategies, including random immunization, acquaintance immunization and targeted immunization. In addition, we give some advises for different objects to administer Micro-blog rumors, aimed to reduce the social impact of Internet rumors.