| ObjectiveAccording to the responsibility attribution of public health emergencies,this paper classifies the events,and uses the related technologies and methods of text mining to restore the occurrence and development process of public opinion of public health emergencies through the comment data of netizens,revealing the development of different types of events The general law and specificity of the dissemination,and provide public opinion guidance countermeasures and suggestions based on the analysis results,and provide auxiliary decision support and basis for the relevant management departments in public opinion research and judgment,analysis and prediction and reasonable guidance.MethodAccording to the different responsibilities of public health emergencies,that is whether there is a clear fault party in the development process of the event,the event can be divided into human fault type event and non-human fault type event.Select the representative events in recent years and crawl the relevant comments of the microblog users on the event.In this paper,we use the heat calculation formula to calculate the comment heat in the public opinion cycle,and divide the public opinion cycle of each event.The LDA topic model is used to identify the topics of comments in different stages of the public opinion life cycle,and the discussion topics of netizens are restored by topic keywords.The emotion in DUTIR is combined into five categories: happiness,hate,sadness,fear and neutrality.Based on this,an emoticon dictionary and an online vocabulary dictionary are constructed for microblog environment.A modifier dictionary is constructed based on previous studies.The above vocabulary is used to analyze the emotion of the comments under each topic.Finally,according to the results of the above process,this paper makes a comparative analysis of the two types of events in three aspects: the evolution of public opinion,hot topics and emotions contained in the topics.Result(1)The evolution period of the event is divided by the heat calculation results,and the heat image is presented;(2)To identify the theme of two kinds of events,17 themes are obtained for human negligence events and 9 themes are obtained for non-human negligence events;(3)The overall proportion of all kinds of emotions of the netizens in the event of human negligence is as follows: neutral: 19.4%,happy: 18.9%,hate: 34%,sad: 8.1%,fear: 19.6%;the overall proportion of all kinds of emotions of the netizens in the event of non-human negligence is as follows: neutral: 27.5%,happy: 16.6%,hate: 20.5%,sad: 5.1%,fear: 30.3%.Conclusion(1)The evolution mode of human negligence events is mainly "single peak",while that of non-human negligence events is mainly "multi peak";(2)Compared with non human negligence events,the public opinion of human negligence public health emergencies lasts longer and the growth rate of public opinion is faster;(3)The general rule of the hot topics of public health netizens is as follows: the occurrence of the incident,the cause of concern,the inquiry of the source,the elimination of the impact,the reflection after the event,and the summary of experience;(4)The phenomenon that the number of topics of human negligence events is significantly more than that of non-human negligence events may be related to the conflict mechanism of the two types of events and the different government response behavior;(5)The main emotion of the netizens of human negligence events is evil,and the main emotion of the netizens of non-human negligence events is fear;(6)Through the longitudinal comparative analysis of the two kinds of events,it can be found that there is a certain emotional accumulation phenomenon for the Internet users of human negligence events,that is,the more similar events happen,the higher the proportion of public bad emotions;while for non human negligence events,there is a certain emotional weakening phenomenon,that is,the more incidents happen,the lower the proportion of public fear emotions. |
PDF Full Text Request