| Since the 20 th National Congress of the Communist Party of China(CPC),the CPC Central Committee has attached great importance to the prevention of major security risks,the dissemination of mainstream ideas and public opinion,the construction of a good network ecology and the construction of a risk monitoring and early warning system,and repeatedly emphasized in many outline documents the important instructions of strengthening the construction of cyberspace and scientifically guiding public opinion to prevent all kinds of major risks.At present,the international strategic pattern has undergone profound evolution,and China is facing major risk challenges.Social network public opinion is the embodiment of social conditions and public opinion,on the one hand,it is the expression of public ideology,on the other hand,it is the expression of netizens’ dissatisfaction with the real society and their practical needs.In the absence of scientific guidance and effective response,it will intensify social contradictions and cause the danger of social division.Under the background of all-media era,social network public opinion spreads in multiple platforms concurrently,which brings challenges to public opinion risk identification.Knowledge graph provides an efficient analysis method for multi-platform public opinion knowledge organization,which is conducive to structural analysis and reasoning of public opinion in multi-platform communication characteristics.With its advanced knowledge organization,rich semantic information expression and open interconnection,the construction of multi-platform public opinion topic graph can disassemble and analyze public opinion from different element dimensions,so as to scientifically and comprehensively grasp the overall characteristics of public opinion in multi-platform communication and the specific similarities and differences in different platforms,and put forward specific strategies for multi-platform public opinion risk prevention in social networks.Based on this,this paper takes social network multi-platform public opinion as the research object,which is composed of public opinions of different platforms.Based on the perspective of information ecology theory,this paper adopts the methods of literature research,empirical research,natural language processing,machine learning,knowledge graph,and system science to construct the topic graph of multi-platform public opinion from different dimensions,and based on this,it puts forward the multiplatform risk identification model and risk prevention strategy of public opinion.It contains six sections in this study.Chapter 3,as the theoretical core of this paper,puts forward a multi-platform topic graph analysis model of social network public opinion,determines the elements of public opinion in multi-platform and corresponds to the information ecological elements one by one,and points out that the multi-platform topic graph of public opinion includes three dimensions,multi-platform user role graph,multi-platform community information graph,and multi-platform spatial-temporal feature graph.Secondly,taking the topic of "7.20 Rainstorm in Henan Province" as an empirical case,in Chapters 4,5,and 6,the user role graph,community information graph,and spatial-temporal feature graph of multi-platform are constructed and analyzed according to the hierarchical logic from micro to meso to macro,and the overall characteristics and internal similarities and differences of multi-platform users,information and spatial-temporal features are clarified,which provides empirical support for Chapter 7 and 8.Chapter 7 builds a multi-platform risk identification model of public opinion based on the results of the analysis in Chapters 4 to 6,which provides support for the multi-platform risk prevention countermeasures of public opinion in Chapter 8.Finally,Chapter 8 puts forward specific strategies for public opinion in multi-platform risk prevention based on the analysis conclusions in Chapters 4 to 7,which is the foothold of this paper at the practical level.The details are as follows.Firstly,in Chapter 3 the elements and relationship model of social network public opinion in the multi-platform topic graph is put forward,and the public opinion in multi-platform is divided into four elements including subject,ontology,space-time,and media,and the relationship between public opinion in multi-platforms is analyzed and dismantled.Secondly,based on the theory of information ecology,the information person,information,information environment,and information technology are one by one corresponding to the elements of public opinion in multi-platform,and a topic graph analysis model of social network public opinion in multi-platform is constructed,which is composed of three aspects: multi-platform user role graph,multi-platform community information graph,and multi-platform spatial-temporal feature graph.Finally,based on the results of the graph analysis,a risk identification model of social network public opinion in multi-platform is proposed,which provides theoretical support for multi-platform public opinion risk prevention.Chapter 3 is the core theoretical framework of the full paper,which expounds the construction and analysis logic of the topic graph of social network public opinion in multi-platform,and provides the theoretical main line for the subsequent chapters.Secondly,in Chapter 4 we analyzed the main elements of public opinion in multiplatform based on the information person factor.Taking the user’s topic characteristics,information behavior characteristics,and social network structure characteristics as user role clustering indicators,the multi-platform user roles are clustered and a multiplatform user role graph is constructed.Using the LDA topic model to identify user topics,and obtain the information behavior characteristics of users through mathematical statistics,the social network structure characteristics of users are determined by the Page Rank algorithm and intermediary centrality algorithm.After clustering with the DBSCAN algorithm to get the user role,a user role graph of social network public opinion in multi-platform is constructed.Second,the user role relationship,user role characteristics,user role collaboration,and user role transformation in the multi-platform are analyzed.This chapter analyzes the users in the multi-platform of social network public opinion from the micro level,which lays the foundation for Chapter 5 to analyze the information from the meso level.It provides theoretical support for the construction of a multi-platform risk identification model of social network public opinion in Chapter 7 and the formulation of a risk prevention strategy in Chapter 8.Thirdly,in Chapter 5 we analyzed the ontological elements of public opinion in multi-platform based on information factors.Construct a multi-platform community information graph to analyze the public opinion information held by different communities in multi-platform.The information graph of the multi-platform community consists of three parts,semantic graph of the multi-platform community,emotional expression graph,and topic evolution graph.Using dependency syntax analysis,semantic role labeling,and text semantic triple extraction methods to construct a community semantic graph,also call Baidu AI emotional analysis interface to identify the emotional attributes and build emotional expression graphs;social network analysis and node importance calculation are used to identify the evolution of public opinion topics.This chapter analyzes the information in the multi-platform of social network public opinion from the meso level,which lays the foundation for Chapter 6 to analyze the spatial-temporal environmental characteristics of the multi-platform of social network public opinion from the macro level.It provides theoretical support for the construction of a multi-platform risk identification model of social network public opinion in Chapter 7 and the formulation of a risk prevention strategy in Chapter 8.Then in Chapter 6,we analyzed the temporal and spatial elements of public opinion in multi-platform based on information environment factors.Also analyzes the spatiotemporal environmental characteristics of social network public opinion in multiplatforms from the perspective of spatiotemporal.The multi-platform spatial-temporal network and single-platform spatial-temporal network are constructed by using the knowledge graph tool.Through the calculation of the local clustering coefficient,the compactness of multi-platform spatial-temporal networks is analyzed,and the centrality algorithm and Louvain community discovery algorithm are used to analyze the important cities and spatial communities in different platform spatial-temporal networks.The spatiotemporal scanning statistical method is used to scan the spatiotemporal gathering points of negative public opinion,and the spatial-temporal pattern of negative public opinion gathering is found,and the similarities and differences of spatial-temporal network characteristics and negative public opinion gathering pattern of multi-platform public opinion are summarized as a whole.This chapter analyzes the characteristics of the spatial-temporal environment in public opinion in multi-platform from the macro level,which provides theoretical support for the construction of a multi-platform risk identification model of social network public opinion in Chapter 7 and the formulation of a risk prevention strategy in Chapter 8.With the results from chapters 4,5,and 6,also based on the holistic view of the information ecosystem,Chapter 7 constructs and analyzes the multi-platform risk identification model of social network public opinion,which takes whether the multiplatform risk of public opinion occurs as the identification goal.According to the information ecological factors,the basic factors leading to the occurrence of risks are determined,and puts forward the specific secondary factors contained in the basic factors combined with the analysis results from Chapter 4 to Chapter 6.The Interpretative Structural Modeling(ISM)method is used to determine the relationship path before the factors,and Bayesian Network is used to identify the risks of public opinion in multi-platform.This chapter corresponds to chapters 4 to 6,which provides theoretical support for the formulation of a risk prevention strategy in chapter 8.At last,in Chapter 8 we summarize the overview,prevention objectives,and strategy system of multi-platform risk prevention of social network public opinion.Based on the analysis results in Chapters 4,5,6,and 7,the multi-platform risk prevention strategy system of social network public opinion is constructed.Finally,the multi-platform risk prevention strategy of social network public opinion is proposed.The prevention system is mainly composed of four dimensions,public opinion users in multi-platform,public opinion information in multi-platform,public opinion environment in multi-platform,and public opinion technology in multi-platform.The corresponding risk prevention strategy of social network public opinion multi-platform is further proposed.This chapter is the practical landing of the full study.At the theoretical level,this paper analyzes and divides the elements of multiplatform public opinion on social networks,and proposes the construction and analysis model of multi-platform topic graph on social networks,which provides theoretical support for the construction and analysis of multi-platform topic graph on social networks.The multi-platform of public opinion is disassembled and analyzed from three thematic dimensions,user roles,community information and spatial-temporal characteristics in multi-platform.On this basis,the integration of the topic graph analysis results and the specific practice of public opinion risk identification is realized,the multi-platform risk identification model of public opinion is constructed from the holistic perspective of the information ecosystem,and the application of knowledge graph and information ecology theory in the research field of social network multiplatform public opinion is deepened.At the same time,the theoretical model and empirical analysis results lays a theoretical foundation for the proposal of multiplatform risk prevention countermeasures for public opinion.At the practical level,empirical analysis is carried out based on typical popular public opinion topics,and the overall rules and specific similarities and differences of multi-platform user role identification,multi-platform community information semantics and topic evolution,as well as the spatial-temporal environment characteristics of multi-platform public opinion were determined.It provides theoretical support for social network public opinion multi-platform risk prevention from the aspects of public opinion multiplatform users,public opinion multi-platform information,public opinion multiplatform environment and public opinion multi-platform technology,and provides scientific guidance for relevant departments to carry out specific practice of preventing major risks in China. |
PDF Full Text Request