The Empirical Analysis And Application Of The Evolution Mechanism About Micro-blog Network

In recent years,with the rapid development of online social networks(OSNs) like microblogging systems, which play increasingly important roles in people’s work and life, people are used to getting information or expressing views through them. It is beneficial to promote the healthy development of microblogging networks or helpul to monitoring and controlling the spreading of information in micro-blog if the depth understanding of evolutionary mechanisms about micro-blog systems is achieved. In this thesis, Sina Micro-blog and Tencent Micro-blog are taken as research platform and the structural features, the evolutionary mechanisms and the information dissemination mechanisms are mainly focused on in the research. The main contributions of the thesis are as follows:(1)First of all, the network datas of Sina Weibo and Tencent Weibo are getten and the empirical analysis of structural characteristics and changes of them as well are discussed. Based on real information of networks’ evolution, the thesis shows the influence of the structure attributes and the distance between the nodes on networks’ evolution. The results imply that the two systems are with similar characteristics, such as small world property, two-stage scale-free distribution on nodes’ in-degree, without assortativity and low reciprocity and so on. Besides, the nodes’ reciprocal rates varies with the degree. All these macro features and laws are not changes greatly over time. In addition, when it comes to the network evolution, the nodes with higher degrees can be more likely to add edges. At the same time, compared with node’s degree, the clustering coefficient or the distance between two nodes plays different roles on it.(2)Taking the theory of link prediction as a research platform, the thesis discusses the relationship of degree or clustering coefficient with the ability of that if one node is more likely to add edges.The method is different from the empirical analysis. Based on the result of the study, a new link prediction index is proposed. The result shows that the conclusion based on the theory of link prediction is the similar to the empirical analysis, but the former method is easier and universal. The new link prediction index also works out well not only on micro-blog network but also on some other online networks.(3)According to the analysis of information dissemination mechanism in micro-blog network, the role of the reciprocity edges in spreading process is analyzed and the information propagation model is improved form SIR model. Besides, based on the assume that the mutually edges can effect the activity of the nodes, a new algorithm is proposed for finding influential nodes. The research shows that the speed and scope of the information dissemination can be improved greatly by mutually edges. In addition, the new ranking algorithm which considers the factor of mutually edges performs better than PageRank algorithm and LeaderRank algorithm for that the nodes that the new algorithm finds can spread information faster and wider.