The Regional Pattern and Hierarchical Tendencies of Service-Oriented Tourist City Network: A Connectivity Analysis of 63 Cities in China

[Background] Previous research achievements of the service-oriented tourist city network have often focused on the analysis of its geographical distribution and service role of the important cities instead of the connections and hierarchical tendencies between the whole city in a large region.[Method]Using big data approaches on the regional connections of 38 tourism organizations including famous hotels, air passenger transport, tourism service agencies across 63 most important tourist cities in China. Fuzzy c-means clustering analysis is used to define 8 city arena clusters. [Results]According to the distributions of connectivity between 63 cities, these eight clusters play different service functional roles in the urban tourism network at four hierarchies. With their “center-edge” memberships, these arena clusters are formed by the interweaving process of regional and hierarchical tourism service connections. The results include the analysis of the various service-oriented tourist city in China and point out the geography “gap” faced by network. [Conclusion] Service-oriented tourist cities need to find their hierarchies and positioning in the network scientifically to avoid blind development, to make regional urban tourism sustainable development.


25
One of most important academic viewpoints in Castells' theory of the network society relates to 26 such practice phenomenon that important tourist cities across some specific large regions are used 27 by different capitals as "basing points" in the connection network of service and production [1]. The 28 resulting connections make it necessary to promote the development of important cities and arrange 29 some service-oriented tourist cities into a connected network hierarchy. However, the lack of 30 theoretical agreement on the defining characteristics of service-oriented tourist cities in the network 31 that perform their important tourism service functions has resulted in scientific taxonomies[2], may 32 usually limited to focus on the highest hierarchical cities [3]. Except for the lack of undisputed 33 definition of service-oriented tourist city itself, the main reason for these eclectic approaches are the the tourism relationship matrix within the region", as The Travel & Tourism Competitiveness Report 2017 [22] shared with us. Therefore, the relevant discussion focuses on the "regional urban tourism never been sufficient analysis and discussed on this aspect in the literature for service-oriented tourist in important world cities [17], unable to cope with the increasing pressure of regional city structural 153 changes and innovation of tourism service products, more and more cities may lose their attraction 154 gradually, and their tourism influence within the region tourism network may decline. The important 155 point is that these tourism services are an indispensable factor of city services that have their own 156 growth potential. Compared with other fields of urban tourism service sector growth, this rapid 157 growth is also the result of the interacting results of demand derived from other sectors [27]. The 158 reason is that tourism service organizations in these cities will benefit greatly immensely from 159 advances in communication technology and information virtualization technology, which will enable 160 them to broaden the spatial scope of tourism services. For example, tourism organizations are 161 generally related to the characteristics of tourism demand groups in specific cities, such as "air 162 tourism organization", "tourism hotel management alliance", etc., but under the conditions of 163 contemporary globalization, due to China's huge tourist scale advantage, it can easily realize "small 164 profit and quick turnover". Some city's tourism organizations choose to implement various chain 165 service alliance strategies in China, there are many tourism subsidiaries, which rely on standardized 166 services to gain competitive advantage [28]. Based on these observations, we focus on the tourism 167 service process between various cities enables us to test such a proposition that there is a complex 168 tourism service, which has distinct characteristics of geographical location and regional patterns.

169
This kind of tourism service needs the cooperation and convenience brought by the distribution 170 of various tourism service organizations. Therefore, the headquarters of tourism service organization 171 may bring more "tourism service value" than the subsidiaries. However, headquarters of tourism 172 service organization usually requires the service-oriented tourist city to have a better location in the 173 urban tourism network [29]. In this context, different headquarters of tourism service organization 174 tend to aggregated distribute in the large cities with perfect urban functions and having a wide range 175 of influences, so these cities are often at the highest hierarchical tendencies in the tourism service 176 network, and they have a leading role in surrounding cities [18]. The higher the hierarchy of service-177 oriented tourist city, the more effective it is to neutralize the distance as an impediment of tourism 178 destination decision-making [17]. In fact, for the consideration of location strategy and tourism 179 externality of high-hierarchy tourist cities, there will be different ranks of cities around the leading   of connected data (http://www.lboro.ac.uk/gawc). According to GaWC's collection principles, this 187 paper treats important service-oriented tourist cities as tourism service centers in specific regional 188 network, so we have tried to develop a method to analyze and measure the networks of service-189 oriented tourist city [31]. It will choose the affiliated type (head office or subsidiary) of various advanced tourism organizations in different tourist cities [23]. Then, there must be some tourism connection between the tourism head office and the tourism subsidiary, so that the city can be given a certain score, for example, the city with the head office is recorded as 5 points, while the city with 193 the general subsidiary is recorded as 2 points. Finally, according to the city scores of different advanced tourism organizations in some given city, the scores of several advanced tourism in the urban tourism connection [32]. This method is not to assume that service-oriented tourist cities 198 will form an obvious city hierarchy, but to designate a tourism network in a region, according to the 199 connection between different advanced tourism industries in various cities, the "hierarchical 200 tendencies" can be revealed [6]. Based on previous urban tourism network research, this can bring 201 two advantages: first, advanced tourism industry has widely representative. Their data are relatively 202 easy to obtain, due to various connections and exchanges between various cities, the data needed is 203 quite huge, this can solve the problem of research data deficiency to the greatest extent; and second, 204 the urban tourism network is to bring a very large number of service-oriented tourist cities into 205 the connection network for analysis. The analysis of the relationship between multiple cities in the 206 whole region is closer to reality, so we can have an overall understanding of the city connection in 207 the tourism region. However, before we describe these results in detail, this empirical principle needs 208 to spelt out the conceptual problems. 209 2.6. Conceptualization：Service-oriented tourist cities as regional centers

210
The concept of important service-oriented tourist city as a regional center has been elaborated 211 in GaWC research [6]. Service-oriented tourist city can be regarded as the center of a regional tourism 212 service network, and various types of tourism service organizations (headquarters or subsidiaries) in 213 network focus on providing services for tourists with different travel purposes and consumption 214 levels. So, the tourism connection network can be formally specified as the interlocking network 215 model (INM). The interlocking network has three levels: network level, various cities at different 216 levels connected in regional tourism economy; node level, some cities playing a pivotal role in a 217 specific region; and at sub-node level, including senior tourism organizations providing different 218 specific services for tourists. So, the formations of service-oriented tourist city network are carried 219 out at three levels, through their cooperation with different levels of network, we can provide 220 seamless connection services for tourists across different geographical locations of the whole region.

221
The connectivity of urban tourism network can be formally expressed by the matrix Vij defined 222 by n cities × m tourism organizations, where Vij is the "tourism service value" of city i to tourism 223 organization j. The value of tourism services reflects the importance of a city to the relevant tourism 224 organizations of tourism service network. Therefore, every column denotes a tourism organization's 225 regional layout strategy and every row describes each city's mix of tourism services. These allow for 226 two types of research: the focus column will let us know about tourism organizations; and the focus 227 will inform our knowledge of various tourist cities. In order to achieve research objectives, our 228 empirical analysis concentrates on the latter type of research, understanding the service-oriented 229 tourist city configuration within the relational data.
Where λab, j is the elements interlock connection between city a and city b in terms of tourism 236 organization j defined in term of matrix V, a tourism service value of a tourism organization in a city.

237
These connections can be aggregated into intercity interlocking links: (2) j=1,2,3,……,n; n is the total number of tourism organizations. Every city has such interlock 239 connection with other cities. All the internal connections of a city are aggregated to form the regional 240 tourism network connectivity (C) of the city: for city a across all cities in matrix V. i=1, 2, 3, ……, m, m is the total number of cities.

242
The limiting situation is that a city does not share tourism organizations with any other cities, so 243 all these basic connections are 0 and it has no connectivity. In fact, in the case of large data sets, the 244 connectivity of the urban tourism network may be quite large. To make them easy to manage in the 245 following use, we express city connectivity as the ratio of the largest connectivity calculated in the 246 data, thus creating a scale from 0 to 1. These scores will be used below to represent the hierarchical 247 tendencies in our analysis.    production, the generated relational data value distribution has good credibility.

287
In this empirical data collection, we are faced with the information of some tourism 288 organizations is very detailed and the information of other tourism organizations is much less.

289
Therefore, by designing a relatively simple scoring system to accommodate multifarious information 290 collected, and selecting the measurement items with the same statistical caliber, this can solve the 291 tension of unequal data distribution. Using the six-point scale (0,1,2,3,4,5), two levels can be given 292 easily, the score is zero when there is no specific tourism organization in the city, and cities houses a 293 headquarter of tourism organizations have a score of 5. Therefore, the key point of scoring decision 294 is to allocate the middle four scores (1, 2, 3 and 4) to describe the tourism service value of various 295 cities. It means that three boundaries must be specified for each tourism organization between 1 and 296 2, 2 and 3, 3 and 4 [34].Therefore, the basic strategy of this paper is to obtain a 2-point score from 297 cities that assume having non-tourism headquarters (i.e. sub-organizations), this score 2 represents 298 the "typical" or "normal" service level of a given tourism organization in a city. To determine this 299 normality, we need to have an overall average of the organization's distribution across all tourist 300 cities. But sometimes some travel service organizations do not reach a given "normal" or "typical" 301 level. For example, if a tourism organizations' service is shared by other cities, and its service scope 302 for a single city is actually small, the corresponding tourism organization in that city will be both 303 scored 1. and a tourism organization in the city showing very few (perhaps none) professional 304 tourism services or tourism participants would also score 1. Generally speaking, the boundary 305 between 2 and 3 is based on the size factor of tourism organization service scale, while the boundary super large tourism organization type with many employees will lead to score 3 in this city where the organization is located, while a tourism organization with a regional headquarters will lead to a an m x n matrix data group. therefore, this matrix data group in this paper has 63x38 specific data 313 array with the Vij ranging from 0 to 5.  organizations. The analysis of cities should not be as many as possible, because as the size of the 339 relational data matrix increases (i.e. more cities have been added.), it may become relatively "sparse" 340 (many zero entries), which reduces the reliability of the analysis. So, these cities were selected for 341 their network connectivity is at least one twelfth network connectivity of the highest city.

368
we find that the hierarchical tendencies and regional patterns of the service-oriented tourist city 369 network have a broad diversity, and the connectivity difference between different clusters is obvious,

370
which provides an ideal insightful interpretation for us. In order to clarify the argument, the results

371
of this paper will be expressed in simplified form, and the "nucleus" and "hybrid" members in the 372 cluster will be identified.

373
This new complex hierarchical tendencies and regional geographical patterns of tourist cities are

385
The regional patterns of city hierarchy in the results is reflected in different arenas in Figure 1.
their cities are roughly located in their geographical locations, which also proves the first law of geography "the space closer the cities are, the more likely they are to be connected". Figure 1 shows the is a large-scale regional structure across eight geographical regions in China. Eight clusters are 391 distributed in four arenas with different tourism functions, in addition to the nucleus cities in the first 392 hierarchical arena. The other three arenas also have some nucleus cities with strong regional 393 functions, and there are some relatively independent singular cities in a small scope, some hybrid 394 cities with strong cross regional influence, a small number of near isolates cities in remote areas. In a 395 whole, these four arenas have different tourism functions and undertakes the task of different 396 tourism chain, these mean that the four arenas have relatively clear regional characteristics.

397
Therefore, the results of these regional patterns reflect the strength of city hierarchical tendencies.

402
In order to further interrogate these results required some detailed analyze at the content of the 403 different arenas. According to the above four types of cities in the arena / cluster, the cities in each 404 cluster or arena are defined as four types of tourist cities: the first type is the nucleus city, that is, the 405 core of the cluster is composed of cities with a affiliations degree above than 0.8; the second type is a 406 singular city, and it is also the members of cluster with affiliations are between 0.3 and 0.8, they have 407 no important member-ship with other clusters; The third type is hybrid city refers to the members of 408 this cluster which share the membership with another cluster; The fourth type, near isolates city refers to the city that does not belong to the cluster( because of its affiliation is not as high as 0.3), but it has 410 the highest membership affiliation with the given cluster.

411
From the above discussions, we should focus on the regional patterns, and we order the analysis 412 and discussion through the differences of hierarchical tendencies.    (Table 3)

478
It is closer to Shanghai in terms of tourism economy, but in fact it is directly affected by Nanjing in 479 terms of administrative jurisdiction.
480 Table 5. The major regional service-oriented tourist city and the second hierarchy.

492
Cluster Ⅵ is mainly composed of some traditional tourist cities like Qinhuangdao, Luoyang,

493
Zhangjiajie, and other important city nodes of The Belt and Road Initiative, such as Quanzhou and transit cities. However, the geographical location between them is far away, and the connection between them is not frequent compared to developed cities, but they are important node cities within 499 the scope of each province. Three cluster in the third hierarchical arena all have no near isolates 500 members, but they all have important hybrid members, these clusters share the unique structure of 501 some tourist cities with other clusters, which is a typical feature of the third hierarchical arena. In

530
First, they lack a competitive tourist attraction, some cities at this hierarchy have some 531 traditional tourism resources such as sightseeing and worship, but lack of high-level tourism 532 resources with strong attraction in current tourism services; 533 Second, the city functions are relatively backward and single, and the outsiders lack the necessity 534 of entering the city.We found that the travel services of these lower-level arena members may be 535 more regional, but their service area is not large. The cluster Ⅶ nucleus including Yinchuan,   These findings come from the analysis of the regional tourist city network in China. However, a 549 famous "center-edge" theories are pointed out in the identification of city network arena. Next, we 550 will analyze the regionalization of contemporary tourism flows how to promote the closer 551 connections within cities in the specific region. This may impinge the long established regional "city 552 network system", which is the traditional focus of tourism geographers in analyzing the relationship 553 between various cities. Figure 1 depicts the "center-edge" structure in this network of China, which 554 gives us some special information.  on the first hierarchy, Tianjin and Shijiazhuang are on the third hierarchy), its regional city system 563 has no second-hierarchy city. These clearly shows that, compared with the "vertical" city relationship 564 of the network of tourist cities in North China, the network of other two regions have a more 565 reasonable "horizontal" city relationship. The good cooperation shows that the tourism service 566 connection between various cities in these two regions has enabled most cities of the two regions to 567 develop healthily and fully enjoy the benefits of regional service connection. Although this results Beijing has adopted absorptive policies on the surrounding cities in urban agglomerations, so that the high-quality resources of urban service functions such as transportation, exhibition, education, medical treatment, high-end shopping in the surrounding areas are concentrated in the core cities.

572
The other two regions pay attention to the spatial distribution of urban service functions in order to

608
Thirdly, the most attractive achievement of this paper is that it also lights shines on the erstwhile 609 tourist cities which were easily neglected in the city network. The important viewpoint is that the indicating that their service network system is not limited to a single cluster, and because of their 616 extensive influence, they will participate in multiple tourism service connections. It is worth 617 mentioning that tourist cities such as Lasa and Jilin are in the network as near isolates members, in is geographically far away from cities in those relatively developed regions such as the East China, industrialized city, and the local industry has declined, its urban function has not made significant relative progress in the past ten years.
of tourist city network. It improves the spatial structure theory of tourism geography in a relatively 629 complete large national region, especially complements the "center-edge" theory. In this paper, we 630 supplement the previous exploratory analysis of the service-oriented tourist city network, the main 631 objectives are: on the one hand, the fuzzy spatial dimensions behind the formation of tourist cities 632 are cleared up; on the other hand, try to describe the geographical details of the network, in order to 633 illustrate geographical space is an important factor in the formation of the network.

634
The research specifies tourist cities as a city interlocking networks that we apply a regional 635 holistic analysis of them in China. Our results obviously reveal the differences of connectivity and 636 geographical distribution within the data, indicating the regional pattern characteristics and 637 hierarchical tendencies distinctions of cities in the network. This hierarchical tendency and regional 638 patterns show three conclusions of regional tourism connection network.

645
we find that there is an obvious phenomenon in the interaction between hierarchical tendency and 646 regional pattern, that is, clusters with low average connectivity are more restricted by regional forces 647 in special region, and they have less influence on cities farther away. This geographical phenomenon 648 further shows that a regional pattern is not only the location space of different clusters, but also  Thirdly, the paper tries to improve our understanding of tourism regionalization by describing number of various tourist cities into a single integrated region to construct an analysis framework of 673 urban tourism connections. The geographical distribution of contemporary tourism service 674 globalization in the regional scope is not an end-product, but an important part in a series of 675 continuous processes. This means that as the connectivity of the service-oriented tourist city network 676 intensifies, the gaps had identified in the network will be filled up in the next few years. On the other 677 hand, with the development of regional growth pole at the national level, the regional service