| With the accelerated urbanization,China’s urbanization rate has exceeded 60%.Facing the problems brought by inefficient management,lagging planning,resource shortage,poor environment,traffic congestion,poor quality of life,and homogenization of urban landscape in the process of urban development,how to build human-oriented city,smart city,and improve urban quality has become an important issue in the construction of new towns in China.The emergence of big data and deep learning technology provides new opportunities and technical support for urban researchers to build city with taking human beings as the essential and improve the perception of cities and society.Compared with traditional urban research data such as census data,survey reports and questionnaires,social media data,which containing users’ geographic location and activity information,records spatiotemporal behavioral information of a large number of people.With wide spatial coverage,fast update speed and low collection cost,social media data is increasingly becoming a new method for future urban research.Natural language processing takes massive text data as the research object,reflecting the location,social relations,subjective perception and other information of individual users.Through natural language processing,researches can query,retrieve and data mining information.It is very important for urban planning and management to research the emotional perception evaluation of urban activity space.This study took Shanghai as the study area which involving 2089,621 Micro Blog sign-in data and 1,521 point of interest obtained from October 7,2019 to October 6,2020 via Python program.Firstly,based on the spatial analysis method,the spatiotemporal characteristics of Shanghai residents’ activities were analyzed.Secondly,using Bidirectional Encoder Representations from Transformers,which was able to break through the simple positive and negative binary classification and classify emotions into six categories(joy,affection,distress,angry,disgust and null),obtained the spatial distribution of emotional perception in Shanghai.Thirdly,part-of-speech analysis was applied to the comments extracted to investigate the cause of the emotional perception.The results are as follows:The overall intra-year changes in the number of sign-in in Shanghai shows the distribution characteristics of "single peak,multiple ridge",with an obvious "electrocardiogram" shape,and the monthly sign-in flow shows a degree of "inverted pyramid".The variation of daytime scale shows a "spoon" shape.At the same time,the time when the sign-in traffic fluctuated sharply coincided with the important time node of COVID-19.The check-in flow in daytime is significantly higher than that in night.And the number of check-in significantly is higher on non-working days than on working days.The traditional routine still dominates residents’ check-in activities,and the correspondence between time and activity content still exists.The spatial distribution of check-in activities is concentrated,with obvious differences in rank.The number of check-in POI and traffic volume both decrease outward from the center of the central city.The popular check-in activity spaces are mainly historical and cultural blocks,characteristic commercial streets,transportation hubs and colleges.Residents have a more positive evaluation of Shanghai’s urban space,with 35.61%positive emotions and 19.7% negative emotions.In terms of time dimension,people are more "distress" at night,with the beginning of human activities,positive emotions gradually increased during the day.From the spatial dimension,macroscopically,there is a concentration of positive and negative emotions in the city center,"love it and hate it".The positive emotions are distributed in a row,mostly in the inner ring area,while the negative emotions are distributed sporadically in the suburban areas,with little regularity.With the distance increase from the city center,the percentage of positive emotion is lower.Microscopically,people’s emotions are highly correlated with the types of activity locations.Mercedes-Benz Arena,Shanghai Disneyland,Hongkou Football Stadium,Global Harbor,SML Center have a higher percentage positive emotion.While activity spaces with a higher percentage of negative emotions are mostly stations and university locations.The top five in order are Shanghai Railway Station,Shanghai Jiao Tong University,East China University of Science and Technology,Donghua University,Hongqiao Integrated Transport Project.People’s emotions are highly related to the types of activity space.Sports and performing arts center as well as theme park shape people’s "joy" and "affection" emotional spaces.While people in Colleges and office experienced "anger" and "disgust".The "distress" atmosphere is strongest in transportation hubs.It is necessary to further improve the positive emotional value of transportation hubs,which can be achieved rationalizing the behavioral routes of users,increasing efficiency,and improving service quality.When dealing with place types with high positive sentiment such as theme parks and,the loyalty of the target group needs to be cultivated through specific activity experiences to avoid the homogeneity of content supply.In order to reduce the negative emotions of college students,we should focus on the reasonable planning of campus space forms,create more spaces for use that can relieve stress,and shape more spaces for enriching campus activities.Paying attention to the protection and revival of historical and cultural landscape areas,and under the principle of protective development and utilization,to reproduce the features,reshape the functions,and create more positive emotional spaces in combination with the unique natural environment of the blocks. |
PDF Full Text Request