Method And Application Of Community Structure Detection For Integrated Development Of Urban Agglomeration

The continuously accelerating process of globalization,knowledge economy,and informatization has greatly promoted the flow of resources between cities.With the expansion and the integrated development of cities within the region,a host of urban agglomerations with various scales have been formed.To enhance comprehensive competitiveness,the regional development plans are implemented,while urban agglomerations are cultivated and developed by many countries.Since the implementation of the reform and opening-up policy,the urbanization of China has been advancing at a high speed and achieved world-renowned achievements.By the end of 2019,the urbanization rate of China has reached 60.60%,which also indicated that the development process of urbanization in China would enter a later mature stage.The decision support for coordinated development of urban agglomerations can be provided by exploring the organizational structure of them.More emphasis has been placed on the hierarchy and scale of cities in the traditional studies of urban system.With the refinement of specialization within the region and the clarification of the role of city,new spatial organization forms and relationships of cities have emerged,which go beyond the research scope of city hierarchy.The city network theory emphasizes the complementarity or similarity of functions between cities,providing a new perspective for studying new types of urban relations.With the development of urban agglomerations,a series of problems have emerged that need to be studied,such as what are the patterns of the organizational structure of cities within the urban agglomeration? What is the changing trend of the organization structure of cities over time? Whether the existing transportation infrastructure and services can support the integrated development of urban agglomerations? How to evaluate the integration of cities within the urban agglomeration from a network perspective?The inconsistent areas can be identified by overlaying the community structure detected over the city network with the boundaries of urban agglomerations delineated by the government.The findings can be used to evaluate whether the cities within the urban agglomerations are well integrated.Focusing on the above-mentioned issues,this paper has carried out community structure detection methods and application research for the integrated development of urban agglomerations,which involved four aspects:(1)the basic structural characteristics over the city networks in China;(2)whether the transportation infrastructure and services can support the integrated development of urban agglomerations;(3)the change patterns of organizational structure within urban agglomerations using geographical awareness network;(4)the community detection over the multiple spatial networks from a comprehensive perspective.This paper aims to explore the organizational structure characteristics of urban agglomerations from three perspectives,that is,real geographic space,virtual cyberspace,and integrated network.The main study contents and findings are as follows:(1)The first content is exploring the basic structural patterns of city networks in China based on multi-variate flow data.The characters and patterns of four networks are explored using indicators and methods from complex network analysis.Two networks are passenger train network and long-distance bus network in real geographic space,while the other two networks are Baidu index network and toponym co-occurrence network in virtual cyberspace.Four main phenomena can be observed: 1)The axial links on the train line network is obvious,and the spatial links among cities within the region is significantly affected by the transportation track.The link strength distribution on the long-distance bus line network presents an agglomeration phenomenon in local regions.2)the degree of city nodes presents a large spatial difference in real geographic space,while the interaction strength between nodes in the virtual space is less constrained by distance;3)the node strength of the above four networks meets a power-law distribution;the node strength has a positive correlation with the size of the city(GDP or population),and has a larger correlation with GDP of a city;4)The node strength located in the eastern regions is significantly larger than that of the western regions.This finding can reflect the lower level of urbanization in the eastern and western regions.Moreover,there are obvious hierarchical characteristics in each urban agglomeration.The patterns presented of the flow data are well understood through the complex network analysis.Moreover,it laid a solid foundation for subsequent studies on community structure detection methods and applications.(2)From the perspective of the real geographic space,a transportation cluster structure detection framework is developed in the second section,which is based on K-shortest path,hierarchical clustering and geographical modularity.The framework including three parts: 1)the K-shortest paths are used to quantify the proximity between cities,which is more in line with people' s travel behaviors;2)a dendrogram is obtained through hierarchical clustering for revealing the structural hierarchies of transportation clusters;and 3)the integration of geo-modularity and hierarchical clustering assures high strength of division of transportation networks.This framework can be applied to various transportation networks.In this section,the proposed framework is applied to the network of passenger trains,the network of long-distance buses,and the combined network of both in mainland China,respectively.The community structure over the combined network is detected.Some isolated cities are identified including Heihe,Ali,and Yushu,etc.Furthermore,by overlapping the detected transportation clusters with the government delineated urban agglomerations,then the transportation clusters namely LX,GP,HC,CP,TCC and WSS,with weak link strength of them are identified.These findings can provide suggestions for transportation planning within the urban agglomerations,thereby promoting the integrated development of them.(3)The third content is the studies on the spatiotemporal change patterns of the city's organization structure within urban agglomerations based on geographical awareness network.Firstly,the concept of ‘geographical awareness network' are proposed,then its characteristics and significance are discussed.With the rapid development of Internet technology and social media,a host of volunteer geographic information(VGI)has provided a rich data for the construction of geographical awareness networks.This study can provide feasible theoretical and methodological support for the subsequent studies.Baidu index was used as the data to build geographical awareness networks in this work.Moreover,spatial data visualization and spatial interaction model are used to explore the patterns of the geographical awareness network.The spectral clustering method is used to detect the community structure over the geographical awareness network.The following conclusions are obtained: 1)users tend to aware of near cities than distant cities.In addition,compared with small cities,large cities are more likely to receive the attention from user from distant cities.2)the first law of geography holds in cyberspace.However,the distance decay effect is weak.3)The first degree of city in the all urban agglomerations has not changed over time.a slice of urban agglomerations including LX,GP,BTH and CP,which presents a single-core structure on the geographical awareness network.The community structure and its changing characteristics over the time are analyzed using geographical awareness network.For urban agglomerations that cannot be well integrated from the perspective of virtual cyberspace,quite a few suggestions are given on human geo-awareness nurturing,infrastructure construction,and urban role positioning,which aim to break the information exchange barriers among cities within the region.Then the flows of information,people,and cargo among cities are promoted,and the interaction strength among cities within the region becomes stronger for integration.(4)Finally,a community structure detection framework by integrating multiple spatial networks is developed in the fourth content.Firstly,the redundancy of multiple spatial networks is measured based on von Neumann's entropy and Pearson correlation analysis,and the redundant networks are removed.Furthermore,multiple spatial networks are integrated using spectral regularization to obtain the community structure from comprehensive perspective.In addition,the framework developed in this study is scalable and can be easily extended to other networks.It can be applied in other fields,for example,gene networks,scientific cooperation networks,and social networks,where a complex system can be considered to be composed of multiple networks.The proposed framework is applied to multiple urban networks collected in this study,and the following conclusions are obtained: 1)Except for Xinjiang,Inner Mongolia,Gansu,Shaanxi,Tibet,and Sichuan,the boundaries of other provinces are completely consistent with the boundaries of the detected communities.To some extent,it shows that the provincial administrative boundary has a certain hindrance to the interaction between cities;2)one community is formed in LZN,BTH,SP,PRD and HBEY urban agglomerations,while two or more communities are formed in remaining urban agglomerations.As shown by the clustering result,the urban agglomerations containing more than two communities mostly cover more than two provinces.For urban agglomerations that cannot be well integrated,decision-making suggestions are given in terms of construction of infrastructure,cultural interaction,integrated spatial structure development of urban agglomerations,institutional guarantee,and resource sharing.The objective is to strengthen collaboration and cooperation among cities within the urban agglomeration and promote the integrated development of them.