| The development of information technology has changed the way of mass communication.The exchange of information and opinions has moved from face-to-face in real life to the virtual space.Individuals transmit information,opinions and behaviors in a peer-to-peer way through social media.This promotes the deep disciplinary fusion of communication,management and computer science.Due to the diversity of social media implementation methods,the universality of participants,and the complexity of communication channels,public opinion communication and evolution on social media are prone to show the characteristics of burstiness,split and complexity.This requires the governments,scholars and experts to face the amount of data and information of social media and scientifically grasp the dissemination law of network events.Then they can analyze the dissemination situation and improve the ability in emergency decision-making.Therefore,the theory of public opinion on social media has become an important research topic.However,most of the research hypotheses of online public opinion researches believe that the social network relationship on the online community has been effective as long as it is established.That is to say,as long as the relevant information or public opinion arrives,it will be transmitted along the social connections in the virtual world.But in reality,due to the influence of users' working,learning,travelling and other social activities,the online social network relationship is in a state of intermittence,and this dynamic relationship composition must have a great effect on the diffusion and evolution of public opinion.Therefore,in view of this social media application scenario,and starting from the temporal characteristics of individuals' online behaviors,this dissertation investigates the temporal features of the online temporal social networks behind social media.Furthermore,taking individual attributes and group social norms into account,the dynamics models on the dissemination and evolution of opinions are studied in the contexts of different types of online temporal social networks.The main research contents and conclusions are as follows:(1)Based on the analysis of the information storage mode of microblog platform and the behavior of users browsing information online,the concept of activity window is proposed to represent the online time distribution mode and limited attention range of microblog users.In this dissertation,the RNE sampling algorithm is used to extract the data collection of the attention relationship network and interaction behavior of 550,000 Twitter users and the statistical analysis results confirm the heterogeneity of the online activity behavior of microblog users.Then,the online activity sequence of each user is mined by the EM-GMM algorithm,and the length of the active window of each user is estimated by weighting.The results show that the statistical characteristics of the activity window accord with the power-law distribution with exponential truncation.This empirical study helps us to understand the online information processing ability of microblog users.(2)According to the characteristics of users' information interaction behavior under the constraints of activity window,an online temporal social network model is constructed,and on this basis,an activity window driven communication model is proposed to carry out public opinion communication mechanism and interaction mode on social media.In this dissertation,social media is not only considerd as a platforms for the public to obtain variouinformation diffusions kinds of information,but also a new communication channel based on human relationship chain and social interaction behavior.They are online temporal social networks theoretically.Based on the above point of view,we use the dataset used in research(1)to build a micro-blog information diffusion network and analysised the structure of its corresponding time accumulation network and time segment network are analysed.Also,we analyze the interaction stability and the burstness in information interaction on it.Then we use the zero-model method to simulate and verify that the activity window is the key driving factor of micro blog information diffusion.This section has certain value for the quantitative understanding of microblog information diffusion base and the dynamic process of information diffusion.(3)On the basis of the above research results,considering the complex contagion of users' online receiving and browsing information behavior,this paper studies the complex contagion model of public opinion on the dynamic network based on the activity window.Under the influence of information technology,the objects,subjects and channels of network public opinion have changed,and complex infection phenomenon is real in network public opinion communication.That is to say,in the process of information diffusion,some individuals must obtain the same information from at least two different friends in order to promote their dissemination.In micro blog,users can browse information pushed by different information sources from multiple active windows for many times,and then decide whether to forward or not This is the true representation of complex contagion in public opinion online.Therefore,based on the complex contagion behavior of user's activity window,and considering the user's cognitive ability and threshold of information,the complex contagion model of public opinion propagation is constructed-Through the simulation experiment,the statistical indexes of complex infection scale,blind area scale and simple infection scale on different time-varying structures are compared,which proved that the complex contagion law of public opinion on dynamic network is different from that on static network.The content of this section is helpful to explain the public opinion transmission mechanism on the dynamic media of micro-blog.(4)According to the different situations of social media users' public opinion interaction,the public opinion evolution model on temporal network based on activity window is construced,which pays attention to the new characteristics brought by the limited attention mechanism to the public opinion evolution of social media.Firstly,in view of the public opinion scenarios where online users have multiple views on an event or have different views around multiple events,individual views are represented by vectors.Considering the characteristics of the active window for online users,the base of group view interaction is set as the online temporal social network with active window,and the evolution and measurement of multi-dimensional public opinion are studied.Secondly,in view of the confirmation bias in online user opinion interaction and their limited attention ability,their online participation in opinion interaction occurs at the time of effective information transmission.The base of group opinion interaction is set as online temporal social network with activity,and the evolution and measurement of Deffaunt model are studied.This section has a certain value for understanding the relationship between user behavior and public opinion situation on microblog dynamic communication media.In all,large-scale data analysis method,data mining method and simulation experiment method are synthetically used to mine the time characteristics of individual online behavior.Based on online temporal social networks,information diffusion model and opinion dynamic models on dynamic network are constructed,and the law of public opinion communication and evolution on the basis of dynamic communication is revealed. |
PDF Full Text Request