| In today's Internet era background,the status of television,radio,newspapers and other mainstream media have gradually reduced.Online media with easy access,real-time,low-threshold begin to perk up.Social networking media as a network carrier,in the case of the integration of social entertainment,news dissemination,the promotion of information and other elements,has become an integral part of the process of people turned outside a window.It is due to the social network comprehensive,complex structure and other characteristic,the public opinion environment in social networks is more complex compared to the traditional public opinion environment.Also,the individualsin social networks have higher initiative,the information transfering between different social groups are more frequent,and the interaction patterns among individuals are more diverse.As a result,the tradition public opinion research is more difficult to apply the above-mentioned new environment.Hence the need to explore the social network information dissemination law,individual interactive mode,characteristic difference analysis and other issues become increasingly prominent.Research needs from multiple perspectives to analyze user behavior,more detailed exploration the characterization of interaction between individual,such as decision-making behavior under different situations,the level of abstraction of the individual interactions.At the same time,it requires a combination analysis of impact and actual propagation range,starting from a complex interaction of social networking context,and study the relationship between competition and information value of information.In view of this,this paper focus on the information and influence propagation in social network,in-depth studies the information dissemination model,information dissemination interference and competition mode,the user impact analysis and dissemination of influence maximization problems frommulti-disciplinary perspective.This study will not only deepen our understanding of complex networks and influence issues,social network theory rich interactive behavior of individual evolution and dissemination of information,butalso can provide effective help in the complex network theory to solve practical problems.The work of the dissertation is supported by the National Natural Science Foundation of China under Grant 61271308 and 61401015,the Beijing Key Laboratory of Communication and Information Systems and the Key Discipline Project of Beijing Education Commission.Main contributions and innovations of the dissertation are as follows:1.Content reliability is an important attribute of information.However,it is ignored in most study of information dissemination in the social network.The system modeling and analysis for reliability factors are lacking in most analysis.According to the actual interaction situation of social network,the information feedback mechanism and the information diffusion model based on the content reliability factor are proposed in this study.By establishing of the average velocity field model,and Monte Carlo simulationexperiment,we deeply analyze the relationship between the reliability of information and the actual communication results,and compare the differences of reliability influence in different network structures.Simulation results indicated us that the reliability of the information content plays a decisive role in the actual results of the information diffusion.Users suspected that the degree of reliability of the information to the extent of the information dissemination of the scope of propagation speed,transmission threshold,the impact of changes in the transmission cycle is obvious.The degree of doubt about the reliability of information has a significant impact on the range,the speed,the threshold value and the propagation period of information.Also,this impact has a low dependence on network density.Furthermore,the user's attention to the reliability the actual influence of information,and it is far less than the final diffusion range of information.In this study,the reliability features of information content areinvolved into the information diffusion study,which provides a deep understanding of the power of information dissemination,and enrich the theory of complex network.It is an effective help for exploring the diffusion and evolution law of information in social network.2.The interference of derivative information is widely existence in the information dissemination process in social network.The traditional information interference model lacks the microscopic description of the interaction behavior users,only analyzes the dissemination process and the function relation from macroscopic angle.According to the interactive features of social network users,we explore the generation conditions of the derivative information,the coexistence-communication mode,and the two information interaction rules.Based on the multi information model and the theory of social reinforcement,a two element propagation model based on the interference of derivative information is established.For the derivative interference phenomenon,the concepts of the advanced-disturbed state and the later-disturbed state are proposed.Also,we purpose social reinforcement rules and users interaction behavior of two elements,and extend the one dimensional transformation user state into the two dimensional transformation user states.In this study,the two element information propagation law based on the derivative information is analyzed,and the comparative analysis is carried out in the regular network and random network.The experimental results show that the first interference isthe strongest one to users in the process of receiving information.The interference form of derived information is mainly on posterior interference.And the timeliness requirement in regular network is lower than it in random network,which make it easier to get interference in regular network.Research on the analysis and modeling of the phenomenon of the derivative information interference are helpful to understand the multi-information diffusion mode of the social network,which provide new research ideas for the study of network interference.3.Individual influence analysis is the most important part of individual heterogeneity in the theory of information communication.The trade-off between efficiency and accuracy has been the key problem of the individual influence ranking algorithm.However,most of the traditional algorithms are less considering the reliability of the algorithm.In this study,we propose a new user influence ranking indicator based the social network topology,which named Iterative Equalization Weight indicator.The transmission characteristics of the influence is be used,and the concept of node centrality and authority is also involved.The new central node increases the convergence speed and reliability of the algorithm.The comparisonson the advantages and disadvantages between IEW index and the other three kinds of classical algorithms are elaboratedin detail.Then,in the simulationof information diffusion,the real data sets are be used to verify our theoretical speculation and analysis.Validation results show that IEW algorithm obtained a high accuracy by sacrificing certain computing speed.Although the accuracy of the IEW algorithm and the PageRank algorithm is essentially flat,but the reliability of algorithm is significantly improved.The new algorithm proposed in this study provides a new idea for mining efficient,reliable and accurate evaluation algorithm of social network node influence,which has a theoretical support for in-depth study of the social network influence dissemination of the law.4.Multi node influence maximization problem is a practical problem in combination with the theory of information communication and the analysis of the influence of nodes.Most traditional algorithmsare difficult to adjust the complexity and time consuming of the algorithm according to the actual demandand the scalability of the algorithm is generally low.In this study,by taking full account of the power-law distribution of node degree in social network and combining the advantages and disadvantages of greedy algorithm and heuristic algorithm,we propose a new multi node influence maximization algorithm based on maxneighbor heuristic.In this algorithm,the greedy candidate node set is constructed by random heuristic algorithm,and then the method of calculating the marginal gain of the node is used to estimate the maximum influence node set.The efficiency and precision of the algorithm are successfully realized in our algorithm.The feasibility and accuracy of the algorithm are verified by mathematical deduction and theoretical proof.We also use the Monte Carlo simulation experiment with real data sets to compare advantages and disadvantages between MNH algorithm and other three classical algorithms.Analysis results show that although the results of MNH algorithm has a more obvious volatility,MNH algorithm has a higher fitness and certain advantages in the comprehensive analysis of the average time consuming and precision of the algorithm.This work well combines the advantages of the heuristic algorithm and the greedy algorithm,providing a new idea to solve the problem of maximizing the influence of multi node,and playing a supporting role in the application of the information dissemination theory in social network. |
PDF Full Text Request