Study And Modeling Of Information Dissemination In Online Social Networks Based On User Behavior And Relationships

The rapid development of online social network has become an important platform for people to obtain and share information,greatly changed the of information dissemination of human society,where is also concerned by researchers in recent years.The existing results have achieved some success,but there are still the following problems:1)Due to the lack of actual data,the models of information dissemination are often based on the static network,depending on assumptions that probability of user behavior and information dissemination are relatively static and equal.it is difficult to describe the real phenomenon of information communication in this way.2)Some issues like the specific behavior of users through which information can be transmitted,how to accurately measure the influence of the user and the efficiency of their effort,have not been generally explained.Based on these problems,this paper obtains and makes use of the actual data of Twitter.Based on the research of user behavior characteristics,this paper makes a systematic description and dynamic model of the process of information dissemination in online social networks.The main contents and contributions of this paper are as follows:1)We analyze the characteristics of user behavior in online social networks,and discuss the related factors and evaluation methods.We make use of information data and user data of the Twitter collection to construct the information communication network based on the retransmitting behavior of user,and analyze the star network topology and its layered structure and degree distribution.Based on the amount of information user published,the wait time and time interval of their action,we introduce user activity,user enthusiasm,user persistence as the evaluating indicators of user behavior in influence.It's found that it is difficult to describe information spreading and user influence directly with user activity,but high enthusiasm and persistence are necessary conditions to obtain excellent information spreading influence for user.We also discover the existence of community differences in user behavior characteristics.2)We use the method of community classification to study the community characteristics of user behavior in online social networks and the Information dissemination macroscopically.Based on the Fast-unfolding algorithm,the Twitter information dissemination network is divided into communities,and the community characteristics of user behavior characteristics are verified.Then we tap the user behavior characteristics of community,and find four groups of users in information transmission and dissemination:information creators,information dissemination facilitator,information dissemination supporters and information consumers.We explain the operation mechanism of their communication and the whole information dissemination.3)We propose a dynamic information dissemination model based on information popularity and user relationship,and carry out empirical test and parameter analysis with actual data of Twitter.Different from the traditional static network model or simulation model,We propose a two-layer dynamic network information dissemination model.The information cloud network is constructed by the time varying publishing information.The user relationship network is constructed by who follows who in online social network.Considering the impact of information popularity,timeliness of information and user relations,we use weight parameters to measure the effect of each factor proportion.We calculate the probability of a user at a moment to retransmit an information to realize the dynamic connection between the two-layer network,and describe the process of information dissemination.Based on the facts of Twitter information dissemination,we also test the accuracy and feasibility of the model.Finally,we discuss the optimal weighting parameters,the effect and relationship of each factors in the information dissemination.The result provide a new perspective for the modeling of information dissemination in online social networks.