Research On Information Spreading Models And Identifying Influential Nodes Of Social Networks

With unique advantages,network science has been crisscrossed in many disciplines in recent years.It provides a systematic new perspective to understand the complex systems or things that multiple interacting entities form,and solves many representative problems in various fields,and favored by experts and scholars.In study of network science,the research on the law of information diffusion and the identification of important nodes has always been the core of network science.This thesis briefly introduces some basic concepts,and reviews the classical model of the networks(Chapter 1),discusses the common models of information diffusion,research status,and transmission control strategies(Chapter 2),and elaborates the research results and current status of important node mining(Chapter 3 and 4).Based on the existing research results,this thesis also explores the following.The main contributions of this thesis are as follows:(1)The DHC theorems on the network are generalized to directed weighted networks,and its performance advantages are verified.Then the method is applied to the China Microblog retweet network between cities.We summarized the online media influence of Chinese cities and the mode of information dissemination on them.Finally,this method was also applied to the analysis of the world merchandise trade networks,analyzing the influence and trade types of different countries,and giving trade strategies to enhance influence(Chapter 5).(2)We proposed a new index to quantify influence of a scholars’ scientific research output,Fan-index.The Fan-index can be more fair while improving the accuracy of the quantification,and it also has good forecasting potential.Then this indicator was extended to the network and found its convergence characteristics,and its converged steady-state value,called the F-index,also has a good ability to identify important nodes(Chapter 6).(3)The information propagation theory and the important node mining method based on circle structure are proposed.We broke through the existing network perspective based on ‘star structure’,proposed and constructed the ‘circle structure’which is a new network substructure,and spread matrix which is used to describe the network based on the substructure,and proposed a new fast search algorithm for high order simple loop.Then,we propose a new index,called Loop number,as an important node identification method on network based on the circle structure.Finally,SIR model as an example,we proposed an information propagation model based on circle structure,it aims to solve a wide range of newly emerging information diffusion phenomena and problems on online social networks.The author believes that the circle structure will complement and expand the star structure and bring new vitality to the study of complex networks in the near future(Chapter 7).