Research On Modeling Netizens’ Negative Emotion During Emergencies Based On ACT-R Architecture

With the development of the Internet,social media has become an important platform for people to obtain and share information.The easy spread of emergency events on social media presents new challenges for emergency response.The spread of emergency events on social media platform is accompanied by the spread of netizens’ emotions.Netizens’ negative emotion aggravates the uncontrollability of online emergency events,and even brings catastrophic consequences to society’s culture,politics,and economic development.The ignorance of the diffusion of netizens’ negative emotion by the subjects involved in emergency events will exacerbate the situation,and also in many cases,the subjects’ responding strategies fail to mediate the negative emotion of netizens,and intensify their emotional reactions.Therefore,exploring the mechanism for netizens’ negative emotion generation process has become an urgent problem for the emergency management of the emergency events.Exploring netizens’ negative emotion is a hot topic in the field of data analysis,emergency management,etc.The current researches on netizens’ negative emotion mainly focus on emotion classification and the emotion modeling.The classification methods of emotion analysis include machine learning-based methods,sentiment lexicon-based methods,and a hybrid method combining machine learning and sentiment lexicons to analyze the objective information of the text.This approach lacks the analysis of the cognitive characteristics of netizens.Netizens’ negative emotion modeling mainly exlplores the mechanism of negative emotion and modeling of evolution process of negative emtion.The former on one hand explores the factors that will influence netizens’ negative emotion through questionnaire surveys and text data analysis,on the other hand,tries to explore how to mediate netizens’ negative emotion from the perspectives of cognitive evaluation and emergency communication.The latter models physical process of netizens’ negative emotion through cellular automaton and infectious disease models.Although some studies have given some cognitive attributes to netizens,they has not formed a unified research system and lacks exploration of the cognitive evaluation process of the evolution of netizens’ negative emotion in emergencies.In summary,the urgent demand of society and researches on netizens’ negative emotion during emergencies both require further exploration on netizens’ negative emotion.Therefore,this research built a framework based on the ACT-R cognitive architecture for the study of netizens’ negative emotion during emergencies,and formed a research paradigm that integrated netizens’ negative emotion recognition,negative emotion evolution modeling,and negative emotion evolution model testing together.The main contents of this paper are as follows:(1)This research integrated the cognitive evaluation features and text features together,constructed a negative emotion feature extraction model of netizens emergencies based on the OCC(the theory proposed by Ortony,Clore,and Collins in 1988)theory,extracted the negative emotion features of netizens and the semantic representation of their negative emotion expressions through machine learning methods.This model was validated based on the case data of Chinese and English social media emergencies,the results showed that this method can improve the accuracy of the emotion classification without increasing the complexity of the model;the cognitive focus of netizens during emergency events is constantly changing,the focus of cognitive evaluation is different in different emergency types,the focus of cognitive evaluation is also different in Chineses netizens and English netizens,the main cognitive evaluation of netizens’ negative emotion is negative,and the cognitive evaluation of netizens’ negative emotion is different from the positive emotion and neutral emotion.(2)This research used netizens’ experience and knowledge about a certain emergency and internalized it into the descriptive memory of the ACT-R system framework and built a model of negative emotion evolution process based on the ACT-R memory mechanism.After obtaining the negative emotion characteristics and semantic representation of negative emotion of the netizens,the model was validated based on the case data of Chinese and English social media emergencies.The research results showed that this method can clarify the cognitive evaluation process of netizens’ negative emotion;negative stimuli makes it easier for netizens to retrieve negative memories,and also makes it is easier for netizens to triger negative emotion;negative memories are more likely to triger negative emotion;the negative emotion is mainly caused by the negative stimuli and negative memory.(3)This research used model checking methods to test netizens’ negative emotion evolution model.This method was verified based on the case data of Chinese and English social media emergencies.The results showed that model checking is an effective time-series data checking method,which can find problems of the model and help to make it better.(4)In terms of practical application,based on the theoretical research and application of the models in case data,this study put forward suggestions on how to respond to social media emergencies from the perspectives of the cognitive evaluation characteristics of netizens’ negative emotion,the negative emotion evolution process and the emotion state transfer.The main innovations of this study are as follows:(1)This study deepens the theoretical foundation of research on the negative emotions of netizens in emergencies.It introduces OCC emotion theory into the study of text sentiment recognition of netizens,and analyzes the Weibo/Tweet text,which is the carrier of netizens’ emotion expression,from three dimensions: event results,behavioral standards,and object attributes,based on the structural characteristics of OCC emotion theory.This study deepens the theoretical foundation for research on netizens’ negative emotion recognition in emergencies.It also introduces the emotion modal model in the field of psychology into the study of the evolution process of netizens’ negative emotion in emergencies,and deeply explains the internal mechanism of netizens’ cognitive evaluation based on their own knowledge and experience after receiving external stimuli.This study deepens the theoretical foundation for research on netizens’ negative emotion modeling in emergencies.(2)This study introduces OCC cognitive emotion theory into the data classification process,and proposes a new method for extracting negative emotion features of netizens in emergencies,namely the sentiment classification method that integrates cognitive evaluation features and text features.This solves the problem of preference error of sentiment annotation.At the same time,this study constructs a cognitive evaluation feature dictionary,which realizes the automatic extraction of cognitive evaluation features of netizens,and reduces the cost of cognitive evaluation feature annotation.It achieves the improvement of sentiment classification accuracy without increasing the complexity of sentiment classification models.(3)This study introduces the cognitive system into the modeling of the evolution of negative emotion of netizens in emergencies.Based on the memory mechanism of the ACT-R framework,it models the process of the evolution of netizens’ negative emotions and it uses the posts of netizens on social media platforms to design the stimuli and knowledge experience in the ACT-R framework system,and fully considers the cognitive evaluation process of netizens in the evolution of negative emotion of netizens in emergencies.It realizes the construction of the cognitive process model of the evolution of netizens’ negative emotions.(4)This study uses the Model Checking method to analyze the dynamic changes of netizens’ emotions in real time during the development and evolution of emergencies.Based on the emotional state transition of netizens in the development and evolution of emergencies,it verifies and corrects the emotion evolution model,and provides new ideas for the verification and correction of models under the background of big data.