Structural Evolution and Community Detection of China Rail Transit Route Network

Round 1

Reviewer 1 Report

This study explored the structural evolution of Chinese railway route networks based on longitudinal data. Overall, the idea is interesting. However, the research tends to include so many analyses and discussion in one paper without a detailed literature review or the method clarification and validation. Besides, the contribution of the work is not clear due to the lack of review. I would like to see the updated version addressing the following points.

 

Here are several main points that need to be considered in the revision.

1)Introduction. pp.2-3. The knowledge gaps or the research motivation is not clearly presented. Could you please clarify it?

2)Before diving into the data and methods, a review section is necessary. Alternatively, a more detailed review of previous work on this topic can be added in the introduction. 

3) Data collection: “on May 30, 2013, December 26, 2016, July 20, 2019 and April 19, 2022 in different time periods.” How were these dates chosen and what are the specific time periods? 

Do these data include both passenger and freight transport? Will these two types of data collection result in similar data formats? 

What is “Jipin schedule /JPSKB.exe “ ?

Add the AMAP API  & 12306 website addresses.

What does D indicate in Table 1. Similarly, other variables that are not self-evident variables in the table need to be explained.

4) Methods: the selection of BGLL is not convincing. Please explain. What are the advantages of this algorithm compared to other community detection algorithm? I would suggest use several commonly used algorithms and selected the best one with highest modularity. See “Discovering the evolution of urban structure using smart card data: The case of London”

 

“The network construction method of CRTRN is applied and accepted by many scholars” What do you mean here?

 

Similar to community detection, the rationale of adopting SDE is not well clarified. A review of previous studies on relevant topics will help.

 

 

How will you estimate the network robustness? More details need to be added.

 

 

5) all the variables shown in table 2 need to be presented in method part.  Those abbreviations in Table 2 are missing in the text.

 

-table 3 is very hard to read. 

 

-Why the robustness analysis only includes 2009 and 2022 without other years? Again, why total degree could be a useful indicator to show robustness?   What does five strategies indicate? E.g., which one is random, and which one is centrality-based?

 

Specific:

p1, “due to COVID-19, public transport conditions in China were poor” what do you mean by condition is poor?

P2. Could you please clarify how the structure of railway would help to save energy and reduce emissions? 

Author Response

Q: This study explored the structural evolution of Chinese railway route networks based on longitudinal data. Overall, the idea is interesting. However, the research tends to include so many analyses and discussion in one paper without a detailed literature review or the method clarification and validation. Besides, the contribution of the work is not clear due to the lack of review. I would like to see the updated version addressing the following points.

Response: Thank you very much for your great comments and revise decision. In the following we have made corresponding replies based on your review comments.

 

Q: Here are several main points that need to be considered in the revision.

1) Introduction. pp.2-3. The knowledge gaps or the research motivation are not clearly presented. Could you please clarify it?

Response: Thank you very much for your valuable suggestions. We adjusted and rewrote the literature review section, and divided our writing motivation into two parts. As “The primary research motivation of this paper is how to improve the partial or overall performance of China rail transit route network, to strengthen the connection between different regions, and form corresponding communities to resist the impact of sudden or long-term external factors. Because the characteristics of the railway network construction of long-term and difficult to change, the research on the optimization of China rail transit route network becomes more realistic significance and long-term social and economic value. Nevertheless, most of the current studies focus on the high-speed rail transit network structure, but these studies ignore the complementary role of other rail transit modes, and the description of the comprehensive rail transit network structure is not inclusive. For example, Feng [53] studied the complexity and structural fragility of China's high-speed rail network and found that its structural change was not obvious from 2015 to 2020, but its robustness was strengthened from 2020 to 2030. Later Jiao [54] studied the robustness of China's high-speed rail network based on the weighted network efficiency metric. Or focus on the impact of rail transit network structure on the surrounding areas. For example, Wang [55] simulated the changes in regional territory, population accessibility and spatial equity brought by the development of high-speed railway network in the Yangtze River Delta of China, thereby defining the impact of high-speed railway on the development of different cities. Wang [56] explored the influence of high-speed railway construction on accessibility of surrounding areas on cities level. Wang [57] believes that high-speed rail has led to the redistribution and transformation of China's tourism market, resulting in greater market competition and the reallocation of urban tourism hot spots. Li [58] based on these studies, argues that high-speed rail increases the mobility of production factors and may also lead to the redistribution of industrial enterprises and their pollution emissions among cities. However, there are few studies on the analysis, community division and robustness of China's integrated railway network [59, 60], and most of studies are based on algorithm or route optimization [61, 62], therefore, this research has great potential and application scenarios. Such as Lu [63] used the widely used community detection algorithm, to study the China’s railway network properties on station level, and found that the number of communities and average distance between community centers are both decreasing. Zhang [64] abstract different forms of railway lines into different network layers, and use multi-layer transportation network model to study the changes of rail transit operating table in different years, there have some discussion on network communities but still lacking.

The secondary research motivation of this paper is how to analyze the railway network in more details. Most of the current studies are based on the network structure at the prefecture or provincial level, limited by data and research accuracy, there are few detailed studies on the station level. For example, Guo [65] studied 20 key cities in the Yangtze River Delta region of China, constructed a combined network of urban high-speed rail network and urban economic network, and proposed the corresponding interactive research framework. Few scholars like Li [66] based on these 1,737 high-speed trains and their stopped stations, transformed the high-speed rail network into a multi-layer network that takes into account both running time and passenger flow, and a corresponding method to comprehensively evaluate the robustness of the high-speed rail network is proposed. And Huang [67] based on the timetable and statistical characteristics of China's railway network and the spatio-temporal pattern of more than 2700 stations, found that the distribution of degree and intensity is scale-free, the average path length decreases and the network clustering coefficient increases. However, the detailed research based on the station level is the foundation support to improve the regional railway network capacity and optimize the overall network performance in the future, which has extremely important research significance.”

 

Q: 2) Before diving into the data and methods, a review section is necessary. Alternatively, a more detailed review of previous work on this topic can be added in the introduction.

Response: Thank you very much for your valuable suggestions. We added a part of literature review during the revision of question 1. It is believed that such modification can make the research status related to this paper have a more clearly display.

 

Q: 3) Data collection: “on May 30, 2013, December 26, 2016, July 20, 2019 and April 19, 2022 in different time periods.” How were these dates chosen and what are the specific time periods?

Response: Thank you very much for your question. These dates are determined based on when we crawled the corresponding data in different years. Due to the system update and other problems, it is impossible to climb the corresponding historical data now, so we can only choose to use the already have historical data and collect the latest data of 2022. They are regular work days or weekends, not holidays. At the same time, the time period should be kept as long as three years as possible, so that the changes of the network can be more prominent, but only the gap between 2009 and 2013 is four years.

 

Q: Do these data include both passenger and freight transport? Will these two types of data collection result in similar data formats?

Response: Thank you very much for your questions and suggestions. “It integrates passenger and freight transport information of the whole system, and provides passenger and freight transport business and public information inquiry service for the society and railway customers.” What this sentence wants to show is that the data of this website contains freight transport information and provides corresponding query function of freight transport information, rather than that the data we used in this paper contains freight train number information. Here we make some modifications to this sentence.

 

Q: What is “Jipin schedule /JPSKB.exe “ ?

Response: Thank you very much for your question, this is the name of the application software, which was launched by Beijing Jipin Time Technology Co., LTD on November 25, 2003. We used the database of this software to reflect the rail transit network operations status in 2009.

Q: Add the AMAP API & 12306 website addresses.

Response: Thank you very much for your valuable suggestion. We have added the corresponding website addresses in the revised version.

 

Q: What does D indicate in Table 1. Similarly, other variables that are not self-evident variables in the table need to be explained.

Response: Thank you very much for your great suggestions. We have added a paragraph before Table 1 to illustrate the corresponding notations, the incorrect header in Table 1 was also modified. As “Here in Table 1, No. stand for the name and number of related rail schedules. Type stands for different types of rails. Station order represents the order of the station in different rail routes. The mileage stands for the mileage of the corresponding stations. Price 1 and 2 are different ticket prices in Chinese Yuan.”

 

Q: 4) Methods: the selection of BGLL is not convincing. Please explain. What are the advantages of this algorithm compared to other community detection algorithm? I would suggest use several commonly used algorithms and selected the best one with highest modularity. See “Discovering the evolution of urban structure using smart card data: The case of London”.

Response: Thank you very much for your wonderful suggestions. The reasons why we choose this algorithm are as follows: first, compared with other algorithms, its computational complexity is lower (O(nlogn)), which is very suitable for the calculation of community division in large-scale transportation network. The time complexity of other algorithms such as Girvan-Newman algorithm is O(n(m+n)), FN algorithm is O(Mn+n^2), Kernighan-Lin algorithm is O(n^2) and so on. Second, its modularity value is very high and its value is relatively stable, can see Table 2. Thirdly, the algorithm has been integrated by Gephi and other software, which is easy to be used in the research of traffic network characteristics. We add these reasons and cited related great articles in the research method section.

 

Q: “The network construction method of CRTRN is applied and accepted by many scholars” What do you mean here?

Response: Thank you very much for your valuable suggestions. What we want to express here is that the illustration of rail networks by using graph theory has been accepted by many scholars. We are sorry for the misunderstanding caused here. In the new version, we change it to "Using graph theory, some characteristics of rail transit network can be expressed clearly and simply".

 

Q: Similar to community detection, the rationale of adopting SDE is not well clarified. A review of previous studies on relevant topics will help.

Response: Thank you very much for this great suggestion. We have added a literature review section before the introduction of the SDE. As “Standard deviation ellipse (SDE) is a classical algorithm to measure the distribution and direction, and it is often used to measure the spatial distribution of some research objects like facilities [78], its application in rail transit network is relatively lacking, it is mainly used to analyze the distribution state of the research object itself or the evolutionary relationship between different objects [79]. The change trend of the center of gravity can be seen by adding the comparison of different years [80]. Such as Ge [81] further compared spatiotemporal variation and evolution of transport network of China's megaregions by SDE. Wyatt [82] studied the distribution of employment and housing situation of Personal Rapid Transit in Melbourne based on SDE. In this paper, it is used to measure the distribution characteristics of rail transit network stations in different years, which can clearly see the evolution trend of different years.”

 

Q: How will you estimate the network robustness? More details need to be added.

Response: Thank you very much for your valuable suggestions. Here we added a part in the research method section to show how we measure the network robustness. As “According to the corresponding network structure characteristic indexes, different network disturbance strategies are constructed to detect the different change characteristics of rail transit network structure facing different disturbance strategies, which has important theoretical and practical significance in the transportation network and operation management. MATLAB will be used to measure the change of network performance by removing the stations in the network through simulation algorithm, and then detect the resilience of the rail transit network. Here, we use the change ratio of node degree value before and after the network is disturbed as the measure index of network robustness. The corresponding disturbance strategy selected in this study is mainly determined according to the maximum node degree (disturbance strategies 1), random node degree (disturbance strategies 2), random connecting edge degree (disturbance strategies 3), and static (disturbance strategies 4) and dynamic node betweenness centrality value (disturbance strategies 5).

The procedure of network disturbance is as follows: First according to the measurement of their initial state network structural characteristics and the corresponding indexes, and then use the corresponding strategy to determine the sorting of different stations, and make sure the disturbance order to related stations, remove the stations from the rail transit network. Once again, measure the properties of the corresponding network structure and the corresponding index, then we can have the value of NR. The centrality of the node betweenness can be subdivided into static and dynamic ones. Dynamic means that the nodes with the largest betweenness centrality value are reordered after each disturbance process, and the nodes with the largest betweenness centrality value are selected to continue the disturbance. While the static disturbance strategy is carried out only according to the order of the initial state of network betweenness centrality value. For the random disturbance strategies, after each disturbance process, a random value is generated until the end of the disturbance. The above process is carried out for 100 times, and the average value of 100 times is taken as the final result of the random disturbance.”

 

Q: 5) all the variables shown in table 2 need to be presented in method part.  Those abbreviations in Table 2 are missing in the text.

Response: Thank you very much for your careful review. We added the definition of average degree in the method part, and changed the symbol of network diameter from D to Di to distinguish it from other symbols. So, all the corresponding symbols are defined in the method part.

 

Q: -table 3 is very hard to read.

Response: Thank you very much for your suggestion. Maybe which you suggested is Table 2. We have modified the corresponding header of Table 2, and added the corresponding definitions in the methods section, which we believe will make Table 2 more clear.

 

Q: -Why the robustness analysis only includes 2009 and 2022 without other years? Again, why total degree could be a useful indicator to show robustness?   What does five strategies indicate? E.g., which one is random, and which one is centrality-based?

Response: Thank you very much for your valuable suggestions. Here, we choose 2009 and 2022 to make a clearer comparison and show the changing trend of network resilience. The figures for other years have been compared, but the changes are not so strong, so we didn't show them here.

In fact, the test of network resilience has received much attention and recognition from many scholars, many of them are based on the study of maximum connectivity subgraph, but this measurement method is not very suitable for our study, because we pay more attention to the connectivity of different rail lines. To some extent, the change ratio of the total degree value reflects the connection ability change of the network and the change of the network in the face of different disturbance strategies. It reflects the overall change trend rather than the situation of one node, so it can be used as a measure index of network robustness.

Here we add a lot of content in the research method and result parts to clearly reflect the corresponding disturbance strategies. The corresponding images were modified to clearly indicate the strategies represented by the different curves.

 

Specific:

Q: p1, “due to COVID-19, public transport conditions in China were poor” what do you mean by condition is poor?

Response: Thank you very much for your question. What we want to express here is that due to the impact of the pandemic, the connectivity and operational capacity of the network have been strongly affected. It has been modified in the new version.

 

Q: P2. Could you please clarify how the structure of railway would help to save energy and reduce emissions?

Response: Thank you very much for your question. The more optimized the structure of rail transit network, the higher the transportation speed and operation efficiency of rail transit, the smaller the average shortest path length, the stronger its ability to connect different cities, and the better the accessibility between different cities. This will gradually change people's preferred transport habits as they move around different cities, from car-based transport mode in the past to more energy-efficient and less polluting rail transport mode, which would help to save energy and reduce emissions.

Reviewer 2 Report

This manuscript try to show the evolution of China Rail Transit Route Network. The data from 2009 to 2022 has been taken for the analysis. It is found that China Rail Transit Route Network is gradually expanding following the southwest direction, the distribution of routes is more balanced, the number of network communities is steadily decreasing, and make various regions becoming closely connected. The topic of this manuscript is very interesting. The study is worthy to provide readers a comprehension of public transportation research. Hence, the reviewer considers the article being qualified for publishing in Sustainability Journal.

Author Response

Many thanks for your valuable review comments and acceptance decision. We will try our best to write more relevant articles in the future.

Reviewer 3 Report

This paper takes the evolution of China rail transit route network (CRTRN) from 2009 to 2022 as the research object，and the evolution of the network structure, the change of distribution, the division of network community, and the network robustness are systematically studied. However, some insufficient description in the research should be modified, and some comments that should be considered by the authors for improving the paper are as follow:

 

1.     In line 240, the authors mention” the COVID-19 outbreak has a certain impact on the development of railway lines”, but did not specify the reason. The question of how the COVID-19 outbreak impact the development and evolution of railway lines needs to be detailed discussed.

 

2.     There are many minor issues, mostly related to language use and repeated words, such as “Beijing station” (line 333) and “eastern China” (line 550). I highly recommend checking the errors and doing careful editing of the manuscript.

 

3.     In section 3.6, How to draw the conclusion that the CRTRN system should pay more attention to the stations with higher node degree values according to the interference situation of the top 100 nodes? Justification of this conclusion needs to be further elaborated.

 

4.     The paper evaluates the evolution trend of railway network structure, but the conclusions obtained are not concise and pertinent, and the future applications expected to be supported by this research are not clearly described. I strongly recommend that the author improve the relevant expression.

Author Response

Q: This paper takes the evolution of China rail transit route network (CRTRN) from 2009 to 2022 as the research object，and the evolution of the network structure, the change of distribution, the division of network community, and the network robustness are systematically studied. However, some insufficient description in the research should be modified, and some comments that should be considered by the authors for improving the paper are as follow:

Response: Thanks a lot for your great comments and revise decision. In the following we have made corresponding replies based on your review comments.

Q: 1. In line 240, the authors mention” the COVID-19 outbreak has a certain impact on the development of railway lines”, but did not specify the reason. The question of how the COVID-19 outbreak impact the development and evolution of railway lines needs to be detailed discussed.

Response: Many thanks for this valuable review comment. After this sentence, we added some reasons why the epidemic affected the development of the rail transit network, and some of these reasons have been revealed by the official media. As “The spread of the COVID-19 has led to the suspension of rail lines in affected areas, which has changed the travel habits of some users. For example, there is now a growing fear among local people of the risks associated with long-distance travel. After the epidemic, due to changes in traffic habits or lines opened, the original rail transit lines will be cancelled due to the reduction of operation costs or the decrease of user numbers, this will force traffic management department to think more rationally about the connections between different areas, which will lead to corresponding changes in the structure of rail transit network.”

Q: 2. There are many minor issues, mostly related to language use and repeated words, such as “Beijing station” (line 333) and “eastern China” (line 550). I highly recommend checking the errors and doing careful editing of the manuscript.

Response: Thank you very much for your comments on the revision of the language part of our article. Based on your comments, we have revised and modified the full text to reduce the corresponding language errors.

 

Q: 3. In section 3.6, How to draw the conclusion that the CRTRN system should pay more attention to the stations with higher node degree values according to the interference situation of the top 100 nodes? Justification of this conclusion needs to be further elaborated.

Response: Thank you very much for your question. We make this conclusion mainly based on the response of network structures to different disturbance strategies. Here, we find that the description of the corresponding disturbance strategy is relatively lacking in the research method section, so we add a part of description content. At the same time, the figure is modified to some extent, and the legend is modified to be named after the disturbance strategy. We believe this change will make this part of the conclusion clearer.

Q: 4. The paper evaluates the evolution trend of railway network structure, but the conclusions obtained are not concise and pertinent, and the future applications expected to be supported by this research are not clearly described. I strongly recommend that the author improve the relevant expression.

Response: Thank you very much for your comments on the conclusions. Based on your comments, we have removed some redundant or less important expressions and added some corresponding descriptions.

Round 2

Reviewer 1 Report

Thanks so much for complying with my suggestions. The changes made by the authors have a significant improvements in the paper. Before proceeding to accept the paper, there are some points needs further revises. 

1) I would suggest to put review section in one separate section right after introduction.

2) As replied by the authors, why this study examine the passengers and freight networks together instead of examining them separately?

3) With regards to the robustness analysis, has this study consider both take into account of natural and sequential attacks? The two types of attacks will affect the orderings of top 100 nodes. 

 

 

Author Response

Thanks so much for complying with my suggestions. The changes made by the authors have a significant improvements in the paper. Before proceeding to accept the paper, there are some points needs further revises. 

Q: 1) I would suggest to put review section in one separate section right after introduction.

Response: Thank you very much for your suggestions. We have divided the original first part into two separate parts in the new version, and modified the corresponding sequence number of rest parts.

Q: 2) As replied by the authors, why this study examine the passengers and freight networks together instead of examining them separately?

Response: Thank you very much for your question. In fact, freight networks are not involved in this article, as we only studied passenger networks. The integrated network studied here is an integrated passenger rail transit network with different operating speeds and models. Maybe the description of the website of 12306 in the data section is somehow misleading, so we have decided to delete that sentence.

Q: 3) With regards to the robustness analysis, has this study consider both take into account of natural and sequential attacks? The two types of attacks will affect the orderings of top 100 nodes.

Response: Thank you very much for your question. In fact, both natural and sequential attacks were considered in our study. The research on natural attacks is based on the Random Node Degree (Disturbance Strategies 2) and the Random Connecting Edge (Disturbance Strategies) 3). The research on sequential attacks is based on the Maximum Node Degree (Disturbance Strategies 1), Static (Disturbance Strategies 4) and Dynamic node betweenness centrality value (Disturbance Strategies 5). Those different strategies determined the orderings of top 100 nodes. To a large extent, these attack strategies can reflect the corresponding robustness changes of the railway network in the face of different natural or man-made attacks.

 

Reviewer 3 Report

Thanks for the modifications made by the authors. There are still several problems in the paper. Please consider the following suggestions when revising the paper:

 1) It is recommended to rewrite the abstract of the paper, this part should include the main contributions, innovative methods and quantitative core conclusions of the paper.

2) In the literature review section, it is not recommended to explain the motivations and purpose of the paper in a very mechanical way. We should first state the problem, and then carry out a research review of the existing problems on this basis, and clarify the existing shortcomings and research gaps by summarizing previous studies, and then put forward the research technical route of this paper. Research motivation should be problem driven or demand driven.

3) In Subsection 2.2.3 of the paper, it is suggested to modify the expression of the reasons for method selection, and describe the reasons from the perspective of the characteristics and advantages of the methods instead of just listing different reasons.

4) It can be found that there are many times to express in the way “...is as follows: ", it appears very blunt and mechanical, not quite in line with the norms of academic expression.

5) There are still big problems in the English expression of the paper. A large number of expressions are not accurate and professional, and the academic expression is also worth improving. It is recommended to ask a professional or a third party to complete and detailed English editing and confirmation of the paper.

Author Response

Thanks for the modifications made by the authors. There are still several problems in the paper. Please consider the following suggestions when revising the paper:

Q: 1) It is recommended to rewrite the abstract of the paper, this part should include the main contributions, innovative methods and quantitative core conclusions of the paper.

Response: Thank you very much for your great suggestion. We have rewritten the abstract part based on your comments.

Q: 2) In the literature review section, it is not recommended to explain the motivations and purpose of the paper in a very mechanical way. We should first state the problem, and then carry out a research review of the existing problems on this basis, and clarify the existing shortcomings and research gaps by summarizing previous studies, and then put forward the research technical route of this paper. Research motivation should be problem driven or demand driven.

Response: Thank you very much for your valuable suggestion. It is true that the previous writing style seems a little mechanical. Therefore, we have adjusted the introduction and literature review section in the new version, which makes the logic between different parts clearer, and also highlights our research objectives and motivations.

Q: 3) In Subsection 2.2.3 of the paper, it is suggested to modify the expression of the reasons for method selection, and describe the reasons from the perspective of the characteristics and advantages of the methods instead of just listing different reasons.

Response: Thank you very much for your advice. We have reorganized this section and hope that the new version will meet your requirement.

Q: 4) It can be found that there are many times to express in the way “...is as follows: ", it appears very blunt and mechanical, not quite in line with the norms of academic expression.

Response: Thank you very much for your great suggestion. We have deleted and rewritten the part involved on the basis of ensuring the integrity of the content. We hope that the new version can meet your requirements.

Q: 5) There are still big problems in the English expression of the paper. A large number of expressions are not accurate and professional, and the academic expression is also worth improving. It is recommended to ask a professional or a third party to complete and detailed English editing and confirmation of the paper.

Response: Thank you very much for your advice and we are very sorry for the inconvenience we have caused. We have asked our colleague with many years of experience in English writing and publishing to revise it, and we hope that the new version can meet your requirements.