Research On The Evolution And Guidance Of Online Group Opinion Based On Higher-order Interactive Networks

In the online ecosystem,social media has dramatically expanded the public’s social interaction space,serving as a crucial channel for people to express emotions,opinions,and demands.When a societal event occurs,individuals in the network quickly aggregate and interact,forming group opinion.Simultaneously,these groups’ opinions are disseminated and interact through online media,influencing the formation and direction of online public opinion.Group opinions exist between micro-level individuals and macro-level networks as a bridge between individual opinion and online public opinion.The evolutionary process of group opinions can be seen as a complex system shaped by the interactive dynamics between groups and their structures.In this system,various factors contribute to a highly disordered morphology,making it challenging for existing Opinion Dynamics and the Collective Action Theory to accurately describe how simple individual interactions interact to give rise to complex group phenomena at a holistic level.The evolution of group opinions is a cognitive phenomenon involving complex interactions,and accurately understanding the evolutionary laws of online group opinions is crucial for observing the structural characteristics of online social networks,trends of online public opinion,and promoting the formation of a healthy online ecosystem.Academia has predominantly approached the subject from a micro-level individual perspective,and systematic outcomes in studying cognitive evolution from a group perspective still need to be explored.In light of this,we aim to integrate knowledge from information science,communication,and complex systems.Harnessing the collective wisdom of computational intelligence,the study endeavors to understand the emergence of group opinions triggered by the aggregation of online individuals from an intermediate-scale group perspective following the eruption of online public opinion.Simultaneously,it employs advanced techniques from the forefront of the complex systems field,particularly higher-order interactive networks,to analyze the evolutionary process of online group opinions.This approach seeks to deepen the understanding of online group phenomena.Based on the identified patterns and critical problems during the evolution,we aim to propose differentiated strategies for guiding online group opinion.These strategies aim to address challenges in network social governance,such as difficulties in identifying network entities,complex network structure relationships,and unclear network ecosystem scenarios.Ultimately,the paper aims to enhance the comprehensive networks governance system,providing dataintelligence-driven insights for cyberspace governance.To achieve the research objectives,the paper follows a research logic of “observing online public opinion phenomena → Finding research questions → Constructing theoretical framework →building higher-order networks → Analyzing evolutionary processes → Summarizing research results → Developing guidance strategies.” Starting from the macro-level network public opinion events triggered by the aggregation of micro-online individuals after public opinion events.Utilizing higher-order interactive networks capable of describing interactions among multiple entities,we propose research on the evolution and guidance of online group opinion based on higher-order interactive networks from a meso-group perspective.After conducting a system review of international and domestic research and clarifying relevant concepts and theoretical foundations,the core sections of this paper are Sections 3 to 8.Section 3 provides a comprehensive overview of the evolution mechanism of online group opinion based on higher-order interactive networks.From a theoretical perspective,it explores the diverse interactive attributes of online groups in the context of social public opinion,elucidating the evolution of group opinions.This section establishes a logical foundation for subsequent sections.Section 4 provides a new methodology for analyzing the evolution of online group opinions by constructing higher-order interactive networks,distinguishing itself from previous binary interactions.Sections 5-7,presented in parallel,aim to dissect,from the perspective of high-order interactive networks,the perception,understanding,and judgment of online groups regarding the environment through analyzing opinion formation,sudden change,and prediction.Based on the results of evolutionary analysis,Section 8 proposes targeted strategies for guiding the differentiation of online group opinions.The overall content of the paper is detailed as follows:Section 3,the evolution mechanism of online group opinion based on higher-order interactive networks aims to answer how to establish a theoretical framework from a new perspective and apply new methods to systematically analyze the evolution of online group opinions in the context of social public opinion.This section focuses on articulating the inherent logic of the research content at the theoretical level,constructing a mechanism model for the evolution of online group opinions based on higher-order interactive networks.Building upon the theoretical foundation of defining and analyzing the concept,characteristics,constituent elements,and influencing factors of the evolution of online group opinions,this section,for the first time,explores the higher-order interactive characteristics inherent in the evolution process of online group opinions from a cognitive science perspective.It provides technical support for constructing a high-order interactive network of online group opinion evolution in the following text.Inspired by the three-dimensional dimension of Situation Awareness Theory,namely the perception of environmental cues,the meaningful understanding of integrated information,and the prediction of future states,this section designs the evolution cycle of opinions exhibited by group users over time in the online interactive environment.The cycle is conceptualized as the formation,sudden change,and prediction of online group opinions based on higher-order interactive networks.It represents users’ perception of environmental changes and anticipation of future situations as they comprehend and judge sentiment information in the network interaction environment.Section 4,the construction of higher-order interactive networks for the evolution of online group opinions primarily addresses how to accurately depict the phenomenon of opinion interactions within online groups on social media after their aggregation.This section aims to accurately depict opinion interaction in online social media groups and provide a methodology for analyzing the evolution of online group opinion in the following text.Compared to traditional pairwise interactions,higher-order interactive networks can fully reflect the high-dimensional and complex network connectivity patterns among multiple-element entities,accurately simulating interactive behaviors in social networks.Given this,we first analyze the construction basis of online group higher-order interactive networks from the perspective of social influence and then innovatively crawl public opinion topic data that can characterize user attributes and opinion interaction relationships from the perspective of post-review-reply.A weighted hypergraph-based online group opinion evolution higher-order interactive networks is constructed.The final network topology analysis and visualization results indicate that this network can accurately show the phenomenon of user aggregation and group interaction in network public opinion and can truly present the existence of the network circle effect in social media.It also provides insights into the impact of user attributes on group aggregation and opinion evolution.Section 5,the formation of online group opinions based on higher-order interactive networks primarily addresses how,after the aggregation of groups,the interactive process among group users leads to the formation of online group opinions based on the perception of environmental cues.This section analyzes the first dimension of Situational Awareness Theory.Utilizing the(hypergraph)graph representation learning method in higher-order interactive networks,the section achieves representation learning of opinions in interactive groups,members,and texts.It introduces an adaptive fusion mechanism that identifies group consensus after integrating multiple opinions.The formation of consensus within a group implies the compromise of some users.Based on this assumption,the paper uses the text semantic differences between individual users and group consensus in opinion content.Using a Whole Brain Model,the section categorizes individuals in the process of forming group opinions into four levels: “Adventurous,” “Sensory,” “Prudent,” and “Preservative.” Additionally,it employs an interpretable machine learning method based on SHAP(Shapley Additive ex Planations)to classify and rank the environmental cues that influence the classification results of group opinions,providing a basis for the hierarchical guidance strategy proposed later based on individual differences.Section 6,the sudden change of online group opinions based on higher-order interactive networks primarily addresses how,after group aggregation,group opinions undergo collective catastrophe in interpreting the meaning of integrated information along with the progress of public opinion.This section analyzes the second dimension of Situational Awareness Theory.Group members often absorb new information and adjust their opinions on public opinion events,resulting in sudden changes in group opinion,such as opinion reversal and bifurcation.To analyze the online group opinion catastrophe based on higher-order interactive networks,we first construct the influencing factors of online group opinion catastrophe based on Information Ecology Theory and identify the catastrophe population using a Hypergraph neural network(HGNN)within higher-order interactive networks.In addition,taking the identified population as the object,the sudden change phenomenon in the evolution process of online group opinion was simulated and verified using the Catastrophe Theory and the Cusp Catastrophe Model.Furthermore,to quantify the catastrophe of online groups,this section draws on the concept of resilience index from previous research and constructs an evaluation model for online group opinion catastrophe based on the resilience index.Analyzing the relationship between opinion catastrophe,resilience index,resilience loss,thresholds of online group opinion catastrophe,and the influence of sentiment changes on resilience provides a basis for the resilience guidance strategy proposed based on emotional differences in the following text.Section 7,the prediction of online group opinions based on higher-order interactive networks primarily addresses how to accurately predict the perspectives of online groups at the next moment after group aggregation.This section analyzes the third dimension of Situational Awareness Theory.By constructing an online group cognitive feature set from a Ternary Interaction perspective,the paper utilizes a Structure Deep Clustering Network Based on Hypergraph(HG-SDCN)as a higherorder interactive network deep clustering algorithm to identify groups in the prediction stage of online group opinions.Building on identifying groups in the prediction stage based on higher-order interactive networks,the section preprocesses sequential opinion data with a time step set at 5 minutes of text publishing.It predicts the opinions of online groups based on a Long Short-Term Memory Network(LSTM),verifying the accuracy of the prediction results.This provides a basis for the later proposed push-pull guidance strategy based on aggregation differences.Section 8,focused on online group opinion guidance strategies based on higherorder interactive networks,primarily addresses how to precisely formulate differential guidance strategies for online group opinions under the background of social media using the results of the analysis of online group opinion evolution based on higher-order interactive networks.After proposing a guidance path for online group opinions based on higher-order interactive networks,the paper analyzes the differentiated results of the data analysis process of online group opinion evolution.It formulates differentiated guidance strategies for online group opinions at different stages of evolution.Through the resolution of the research questions mentioned above,the theoretical significance of this paper lies in enriching and improving the theoretical framework of research on group cognition in social media and opinion dynamics.It also expands and deepens the application scenarios of higher-order interactive networks in opinion evolution and public opinion guidance.Regarding practical significance,the research can assist social platforms and regulatory authorities in accurately depicting the networked dissemination patterns of online group opinions and complex decisionmaking modes.It provides scientific guidance for the governance of social media ecosystems and public opinion,serving as a reference for the healthy development of social media and the management of online cultural ecosystems in China.