Social Network Public Opinion Propagation Guidance And Dynamics Analysis

Online social networks(OSNs)offer a wide range of social services that are gradually transforming individual lives and social habits.One of its most frequent services is the information retrieval and hot topics discussion.The user attitudes,variety of opinions,and emotional depiction make OSNs have emerged as the most important platform for public opinion.However,the cyber-violence and group polarization of public opinion result to the unfavorable atmosphere and bring a threat to the peace of the social community.Meanwhile,public opinion also reflect of the social issues and contradictions in the real society.Hence,in order to monitor public opinion and preserve network harmony,it is crucial to conduct deep research and discuss the key components of network public opinion,including online users(subjects of public opinion),information(ontology of public opinion),hot topics(ontology of public opinion)and its opinion dynamics.Although there have been many advances across several academic disciplines in the study of network public opinion,there are still certain issues that need to be solved.Firstly,for user modeling,the present method for modeling online users lacks individualized analysis because it overlooks the interactive process of social influence and the process of user opinion creation.Secondly,for information propagation and guidance,the present information guidance strategies have not had in-depth discussions about the dissemination mode of multi-information competitive scene and time-limited scene,which makes it difficult to support the analysis of concurrent dissemination,and also ignores the influence of user topic preferences and opinion tendencies on the results of opinion guidance.Last but not least,in terms of opinion dynamics perception and prediction,most research have focuses on the current state of public opinions,including its popularity,influenced users,etc.,but neglects to consider the role of emotional content in the dynamic evolution of public opinions.Because of this,current methods are unable to perceive the dynamics of public opinion from both from both popularity and emotional trends.· Aiming at the issue of user modeling,an agent-based modeling of the personalized user is proposed to thoroughly examine the process of user opinion formation in OSNs.Fist of all,the Agent-based user Opinion Formation model(AOF)considers the multi-dimensional influence factors from the topic community,friends,and user self,and a deep fusion method are used to predict user opinions on the topic in the future.Furthermore,in order to construct a dynamic social network environment among agents,a Multi-Agent System for Finegrained Opinion Dynamics analysis(MAS-FOD)is designed and implemented to simulate the interaction influence process between agents.The experimental results based on practical data prove that the AOF can predict the user opinion of the topic more accurately.Meanwhile,the MAS-FOD simulation results show that this method can also provide correctness guarantee for the trend prediction of group opinions.· Aiming at the issue of information propagation and guidance,especially for the competitive scene and time-limited scene of hot topic information propagation,this paper constructs the propagation model and influence maximization algorithms for specified information guidance.Firstly,based on the AOF,the user’s opinion tendency is predicted to form a set of candidate seed users.Severally,with regard to the competitive propagation scene,fully considering the synchronous/asynchronous characteristics of mutually exclusive information cascade,and based on user topic preferences and three-degree influence theory,a multi-information competitive propagation model and the influence maximization algorithm are proposed to provide guidance strategy for the specified opinion of information.Besides,with regard to the time-limited propagation scene,a comprehensive analysis of user online status,topic preference,interaction delay and propagation path is carried out,and a time-sensitive information propagation model and influence maximization algorithm are proposed to provide strategies for the time-limited dissemination of information.The experimental results under practical data show that the model and algorithm proposed in this paper have the optimal influence result,and balance the computational efficiency of the algorithm.Therefore,they can provide the best guidance strategy for specified information or time-sensitive information in the real scenario.· Aiming at the issue of opinion dynamics perception and prediction,this paper provide a Multi-view Aware of Public opinion Prediction(MAPP)for measuring public opinion perception of hot topics in OSNs.The three dimensions of information propagation structure,user content,and time series are used by the model to learn the multi-view characteristics of public opinion at each time window of spread.To forecast the future condition of public opinion,the two aspects of information popularity and emotional tendency are then used.The findings of the practical data-based experiments show that the model can more correctly predict the popularity of public opinion information spread as well as its emotional propensity.Through the above research work,this paper focuses on the three components of public opinion propagation: user,information and public opinion of hot topics,especially for the deep discussion of user modeling,information propagation mode and opinion dynamics analysis,the models and methods for user opinion formation,information dissemination and guidance,and public opinion perception and prediction are solved.Hence,this paper effectively provide technical and theoretical support for network public opinion warning,supervision,guidance and other services.