主管单位:中国科学技术协会
主办单位:中国地理学会
承办单位:华东师范大学

世界地理研究 ›› 2024, Vol. 33 ›› Issue (1): 134-148.DOI: 10.3969/j.issn.1004-9479.2024.01.20220207

• 城市与产业 • 上一篇    

中国城市信息流网络空间结构特征研究

安頔(), 胡映洁, 万勇   

  1. 上海社会科学院应用经济研究所,上海 200020
  • 收稿日期:2022-03-27 修回日期:2022-07-13 出版日期:2024-01-15 发布日期:2024-01-29
  • 作者简介:安頔(1988—),男,博士研究生,主要研究方向为城市与区域规划、城市经济, E-mail: andi.008@163.com
  • 基金资助:
    国家自然科学基金青年科学基金项目(41901199);国家社会科学基金重点项目(19AZS018)

Analysis on characteristics of urban information network structure in China --A method based on denoising and directed network

Di AN(), Yingjie HU, Yong WAN   

  1. Institute of Applied Economics, Shanghai Academy of Social Sciences, Shanghai 200020, China
  • Received:2022-03-27 Revised:2022-07-13 Online:2024-01-15 Published:2024-01-29

摘要:

使用ICEEMDAN方法和有向网络分析方法,对中国336个地级以上行政单元间百度搜索指数进行降噪处理、构建有向与无向加权网络,分析2014—2019年网络层级结构、节点不对称性、空间组织结构等方面的演变特征。发现:①大数据降噪处理具有必要性,降噪处理获得的长期趋势数据能有效揭示网络节点演变特征;②信息流网络具有显著不对称性特征,有向网络分析方法能更好地揭示节点的异质性特征;③网络节点成长路径具有非均衡性,对内、对外联系增长表现出阶段性特征;④信息流网络具有全国范围和城市群、省域范围两种尺度组织形式,分别表现出异配性、无标度网络特征和同配性、小世界网络特征,兼具长距离联系与地理临近的特点。这表明信息流网络兼具中心地和网络化两种理论范式的形态结构特点,体现出虚拟空间与实体空间的关联性。

关键词: 城市网络, 信息流, 百度指数, ICEEMDAN, 有向网络

Abstract:

This research used the newly developed tool of ICEEMDAN (improved complete ensemble empirical mode decomposition with adaptive noise) to decompose the Baidu index among 336 prefectural-level administrative units in China. Then this paper constructed directed networks to compare with undirected networks by analyzing the evolutionary characteristics of urban hierarchy structure, node asymmetry, spatial organization, and structural complexity in information networks between 2014 and 2019. The results show that: ① A large amount of noise interference exits in intercity Baidu search data, and noise reduction processing is necessary. After noise decomposition, the long-term trend data can better reveal the evolution characteristics of nodes. ② Considering the significant asymmetry of the information network, directed network analysis is a more effective method than undirected network analysis to identify the difference between node characteristics and their trends. ③ Node degree centrality shows a disproportional growth path, demonstrating growth dominance shifting from in-degree to out-degree at different stages. ④ Dominant flow analysis sheds light on two spatial organization patterns in the information network. One pattern is on the national scale, with disassortative network and scale-free network features. The other is on urban agglomerations and provincial scales and features assortative network and small-world network properties. The information network organization is a complex process, especially with the differences in long distance and geographic proximity interactions as two patterns indicate. This illustrates the characteristics of the spatial organization of two models, the Center Place Model and the Network Model, as well as the interaction between cyberspace and physical space.

Key words: city network, information flow, Baidu index, ICEEMDAN, directed network