An Assessment Framework Of Green Space Satisfaction Using Social Media Data

Green space has irreplaceable economic and social benefits for the human ecosystem,and plays a series of functions in the process of sustainable urban development such as biodiversity maintenance,urban heat island effect reduction and physical and mental health promotion for human beings.The evaluation of green space satisfaction(GSS)has long been the interest of a wide variety of studies in order to enhance the quality of green spaces for sustainable urbanization.However,the procedure to construct a rapid and effective method for large-scale GSS assessment is still challenging.In the era of big data,new approaches and alternative data sources,especially volunteered textual data can provide large scale and large amounts of information about GSS,but how to effectively interpret this volunteered data is yet fully explored.In the Chinese society where the urbanization process is entering an inflection point,the technical methods for green space evaluation,development and management require further development to correspond to the refined governance of urban space in the context of inventory planning.Using social media data(SMD)about urban parks in Beijing,this study proposed a content analysis framework adopted based on machine learning and word-matching to interpret both overall park satisfaction and satisfaction to individual landscape elements.Several tests with evaluation indexes confirmed that our machine learning approach outperformed four available tools and could automatically and effectively interpret volunteered textual comments into green space satisfaction.Based on the above method framework,this paper takes 50 representative urban parks within Beijing's sixth ring road as the research object,and integrates landscape pattern analysis to obtain the following main conclusions:(1)Visit frequency of parks is related to the characteristics of green space and seasonal changes.The distribution of 50 park' visit amount demonstrates a significant long-tail effect.Visit frequency between parks are different,among which,the popularity of the Summer Palace and Beijing Zoo rank in the first echelon,leading other parks.Pearson correlation coefficient between total area and visit frequency reaches 0.579(p <0.001)which shows significant correlation.But through a simple qualitative analysis,it can be found that the effect of area on the frequency of visit is not decisive.Variables such as accessibility,popularity may also have greater impact on the popularity.Visits frequency of parks in Beijing reaches its peak during spring,summer and early autumn(especially in April,July and October)while reach the bottom in winter.The pattern revealed by social media data not only explains the dynamic changes of tourists' behavior behind the phenomenon,but also provides a reference for traffic detection and facility management of green spaces.(2)The overall satisfaction of case study parks is high,but still exists certain difference between different parks.The satisfaction on the individual scale does not conform to the normal distribution implying that users tend to give positive ratings on social media.The overall satisfaction of all parks also shows similar positive trend However,there are large differences in satisfaction among some parks.The results of qualitative analysis show that landscape elements,especially natural elements,and the iconic landscape of the parks have great impact on park satisfaction.(3)The perception of different landscape elements is quite different and elements have a significant impact on the overall satisfaction of the park.The perceived frequency of elements vary greatly between different parks,especially natural elements and supporting facilities while the distribution of perceived satisfaction is relatively concentrated.Four types of typical parks are identified by clustering the perceived frequencies of three types of elements——natural elements oriented,cultural elements oriented,supporting facility oriented and weak element perception.Among four types of parks,the satisfaction of natural element oriented parks is relatively high while the overall satisfaction of parks with weak perception are relatively low.A further statistical analysis demonstrated that the most influential factors of overall satisfaction were satisfactions to individual landscape elements but not the frequency of landscape elements.The result of linear regression model show that some landscape elements have a greater impact on overall satisfaction,especially vegetation,water among natural elements and recreational facilities.(4)Landscape metrics are able to quantify the spatial structure of green space and directly affect element perception of tourists,thereby further influencing the overall satisfaction of the park.Based on the comparative analysis of subjective perception of the park(overall satisfaction,element perceived frequency and satisfaction)with landscape metric analysis,the results indicate that: the overall satisfaction is affected by structure of landscape elements to certain extent.The structural complexity of tree,water and pavement are strongly correlated with satisfaction and demonstrates a tendency that the higher the complexity,the lower the satisfaction.In terms of element perception,the impact of landscape structure is more obvious on the perception of natural element rather than cultural element and supporting facilities.The statistical analysis proves that complex spatial arrangements of patches rather than total area or area proportion significantly affect visitors' perception of elements and three types of landscape metrics including patch density,shape complexity and irregularity of individual element play major roles.The findings can provide direct guidance for the planning and design of green spaces.for example,at early stage through landscape morphological design with careful index calculation,the proper layout of landscape element could largely increase visitors' perceived probability;The enhancement of cultural element and facilities mainly relies on more ingenious and sophisticated spatial design and construction besides following rules mentioned above.Combing spatial and structural intervention with reasonable design techniques adopted in the process of construction,the satisfaction of green spaces can largely be maximized.This article first proposed a green space satisfaction(GSS)evaluation framework based on social media data.Through case study,the framework confirmed the potential of volunteered comments online as complementary information to traditional surveys in assessing GSS while enhancing our understanding of GSS at a regional scale with comparison of various parks in order to facilitate sustainable policy making of urban green spaces.