The Characteristics And Comparative Analysis Of Changes In Shenzhen’s Emotional Maps Under Novel Coronavirus Pneumonia

At the beginning of 2020,the Covid-19 pandemic,which broke out with people returning home during the Spring Festival,has significantly impacted people’s emotions.Thus,urban construction pursues higher quality and focus more on the perception of “people” in the city.Under this background,this essay explores the relationship between citizens’ emotional perception and the built environment from the perspective of novel coronavirus pneumonia.For these complicated problems,Big Data from the Internet has provided massive data for support.The skill of machine learning has become an efficient approach to summarise its inherent rules.Therefore,this essay applies the emotional information contained in the big data of Weibo text and uses machine learning methods to construct the emotion recognition model based on the built environment both in normal months and the epidemic months.On the one hand,it fills and corrects the areas with an insufficient number of Weibo text in the emotional map.On the other hand,it combines the importance of features output by the emotion recognition model to further sort out and analyze the relationship between the built environment and emotions.This essay aims to have a more comprehensive analysis of urba n emotions’ characteristics,firstly determining that emotions are divided into three dimensions,emotional intensity,motivation,and polarity,through the review and summary of relevant literature.Moreover,emotions are classified into seven types,happ iness,good,anger,sorrow,fear,hate,and surprise.The built environment elements that may affect emotions are divided into 23 sub-elements in five categories,land use,spatial form,road traffic,development space,and public infrastructures.After th at,the Weibo data in Shenzhen from June 2018 to December 2018 and during the pandemic in February 2020 are collected through web crawlers,and each Weibo text is translated into emotional data through semantic analysis.Subsequently,based on the method of building up the relationship between the urban emotions and the built environment before and after the pandemic,a three-scale mesh method combined with two integrated learning algorithms based on the emotional data from normal months and the epidemic months is used to select grids with sufficient Weibo texts.It then constructs an emotion recognition model based on the built environment according to these data,respectively.After checking the validation set’s accuracy,it is found that the recognition model constructed by using a 1000×1000m grid combined with the random forest has higher accuracy for the three emotion dimensions.For the five types of emotions(Fear and surprise are not concluded in the model because of limited data of these emotions),t he recognition model constructed by an 800×800m grid combined with the random forest has higher accuracy.Next,through the application of the corresponding emotion recognition model,girds’ emotions,where the number of Weibo texts is insufficient,are fi lled and corrected.The emotion distribution characteristics of Shenzhen in normal months and epidemic months are analyzed by visualizing the emotion map.The changing features of emotions during the pandemic are calculated based on this basis.In the meantime,the importance of features output by the emotion recognition model is used to analyze the changes in the importance of the built environment during the epidemic and select the built environment elements that have a significant impact on emotions.Then,through the characteristics of the built environment with strong influence corresponding to different emotions,the built environment’s effects on the change of emotions and the reasons for the changes in emotions during the epidemic are analyzed.Finally,corresponding enlightenment is provided for future urban construction of Shenzhen based on the above analysis.