地球信息科学学报 ›› 2018, Vol. 20 ›› Issue (5): 593-601.doi: 10.12082/dqxxkx.2018.180066

• “海上丝绸之路空间数据分析”专辑 • 上一篇    下一篇

大规模航运数据下“一带一路”国家和地区贸易网络分析

孙涛(), 吴琳, 王飞, 王琪, 陈昭, 徐勇军   

  1. 中国科学院计算技术研究所 专项技术研究中心, 北京 100190
  • 出版日期:2018-05-29 发布日期:2018-05-20
  • 作者简介:

    作者简介:孙涛(1993-),男,博士生,主要从事时空数据挖掘研究。E-mail: suntao@ict.ac.cn

  • 基金资助:
    中国科学院重点部署项目(ZDRW-ZS-2016-6)

Analysis on the Trade Networks of the Belt and Road Countries and Regions under Large Scale Shipping Data

SUN Tao*(), WU Lin, WANG Fei, WANG Qi, CHEN Zhao, XU Yongjun   

  1. Special Technology Research Center, Institute of Computing Technology, Chinese Academy of Sciences, Beijing 100190, China
  • Online:2018-05-29 Published:2018-05-20
  • Contact: SUN Tao E-mail:suntao@ict.ac.cn
  • Supported by:
    Key Project of the Chinese Academy of Sciences,No.ZDRW-ZS-2016-6.

摘要:

在“一带一路”沿线的65个国家中,46个国家拥有登记在案的港口,同时海上航运贸易占国际贸易总量的75%以上。为了充分了解“一带一路”沿线国家和地区航运贸易情况,评估国家、区域之间贸易往来关系,本文选取了2016年“一带一路”国家和地区船舶历史运动轨迹,首先基于规则判定的方法挖掘船舶停港事件,并以港口为主要节点,港口间货运往来事件为连接形成“一带一路”国际航运贸易网络。在此基础上,对贸易网络进行如下网络结构分析:① “一带一路”贸易网络基本属性统计,包括网络连通性、度分布、平均最短路径;② 网络节点中心度计算,主要采用Eigenvector Centrality评估分析贸易网中节点中心度;③ 结合社会网络挖掘中社区挖掘的概念,使用Fast Unfolding算法对贸易网络进行社区发现。可以看出,“一带一路”沿线国家和地区贸易往来错综复杂,港口之间呈现小世界网络特性;土耳其、俄罗斯、中国等国的港口影响力靠前;并且形成五大贸易社区,这些社区的分布和地理位置分布基本吻合,但仍然有部分国家受特殊贸易行为的影响,所属社区有所打破区域限制。本文旨在通过航运大数据构建贸易网络,在网络分析基础上,更好地评价节点影响力,更清晰地分析贸易网络结构,为“一带一路”战略更好地实施提供帮助。

关键词: “一带一路”, 贸易网络, 网络分析, 节点中心度, 社区发现

Abstract:

Among the 65 countries along the Belt and Road, 46 countries have registered ports of entry. At the same time, the trade by maritime shipping account for more than 75% of the total international trade. In order to fully understand the shipping trade in the countries and regions along the Belt and Road and assess the trade relations between countries and regions along the Belt and Road, we selected data which depicts the shipping history movements of the countries along the Belt and Road in the year of 2016 for study in this paper. Firstly, based on the method of rule determination, we excavated the Stop-port events of ships. By use of the ports in the countries of the Belt and Road as the main nodes, and the inter-port cargo transactions events as the edges, we have built the Belt and Road international shipping trade network. Based on this, the following network structure analyses of trade networks were conducted: (1) basic attributes analysis of the Belt and Road trade network, including network connectivity, degree distribution and average shortest path; (2) calculation of network node centrality, mainly using Eigenvector Centrality to evaluate the centrality of nodes in the trade network; (3) Using the concept of community mining in social network mining as the reference, and using the Fast Unfolding algorithm to discover the community of trading network. It can be seen that the trade between the countries and regions along the Belt and Road is intricately interwoven. By analyzing the degree distribution of nodes in the trade network, it can be clearly seen that there are small-world networks within the Belt and Road trade network. Further, Turkey, Russia and China are the three most influential counties in terms of the ports influence. By analyzing the results of the community detection, five major trade communities were identified. The distribution of these communities is basically in line with the geographical distribution. However, there are still some countries that are affected by special trade practices and their communities have broken regional restrictions. By building the trading network under large scale ship data, we evaluated the node's influence and analyzed the structure of the trade network more clearly on the basis of network analysis, and we hope this paper can help to better implement the Belt and Road Initiative strategy.

Key words: the Belt and Road, trade network, network analysis, node centrality, community discovery