Research On Measurement Of Information Diffusion Ability Of Node And Information Diffusion Laws In Social Networks

With the development of Web2.0 and the rapid expansion of the scope of Mobile Internet, online social networks have gradually become the main channels of information diffusion and communication. Facts and studies have shown that online social networks play an important role in large-scale information diffusion. Online social networks are social networks of users that forming in the using of computers and networks as the intermediary to realize contraction and social interaction and collaboration. The popularity of online social networks makes study about information diffusion based on social relations have become the focus in the current. This thesis intends to combine the current research status at home and abroad, study the measurement of information diffusion ability of node and niformation diffusion laws in social networks. This study enriched the theory and method of social network analysis and network information diffusion. This study provide a basis for more effective marketing strategies, help enterprises to better realize the advertising fixed-point delivery, for the government’s public opinion surveillance, the prediction and intervention of network diffusion of emergency and group events, and the effective guidance and control of the rumors are all has the vital significance.In this thesis, the research contents and conclusions are as follows:First of all, we propose a centricity measurement in weighted undirected social network for node communication ability based on the ideal of h-index, and give further various extensions, study the measure effect in actual social network, and also analyze the similarities and differences with other node centrality measurements. Numerical simulations and statistical modeling approach are used to contrast the centrality measurement in the identification of node information diffusion ability. The results show that the methods are different with other methods and have the following characteristics: simple and easy to operate, only depend on the network structure, have universality, and can reflect the existing centrality measurements. The presented methods have a strong ability to identify information diffusion ability of node.Second, this thesis extends the research above to directed network, and presents a new method to measure the information diffusion abilitiy of nodes in directed network, and also gives extensions of it on many aspects. Applying this method to citation network to measure the article’s influence, we then propose a method to measure the knowledge diffusion ability of scholars based on the citations relationship of articles. Taking the Q&A community Zhihu for cases, we present a new method to measure user’s ability of spreading high quality informations. Example analysis shows that the proposed method is effective, and thus provides a new method and instrument for measuring the diffusion ability of nodes in the social network, and for identifying the influential spreaders and for the centrality analysis of nodes.Then, we study the correlation between position of nodes in a dynamic social network and its information diffusion ability, and also establish the forecast model of user future information diffusion ability based on the user’s position in the early social networks. Studies have shown that the node occupies the important position in early networks and then are more likely to occupy the important position in the future. Node centralities can be used to predict its information diffusion ability in the future networks.Combining with the message forwarding rules in microblog, using the bayes principle, the information diffusion model is established. The model can depict the changes of spread will of users, describe information diffusion process and the spread will of users and range of known users, and also, it can undertake the effects prediction of information control behavior such as the fixed-point immunity. The simulating research on real network shows that,this model can effectively describe Weibo message forwording rule,reappear the information propogation process, and then provide new method for forecasting the range of information diffusion and range in known state, and for controling of the information diffusion.Finally, from the aspects of information itself, the user behavior, user attributes and social relations and information exchange relations, this thesis analyzes information diffusion rule and influencing factors in online social networks. Study shows that there are rules to follow in information diffusion, the length of the information, the user’s attributes, the user’s behavior and the user’s social relations will have an effect on information diffusion.