Measuring interaction among cities in China: A geographical awareness approach with social media data

Xinyue Ye, Shengwen Li, Qiong Peng

Research output: Contribution to journalArticlepeer-review

9 Scopus citations


Unlike the large body of research on investigating interactions among cities using survey data, the social media-based city interaction study has received much less exploration. Based on geographical studies of social media content in China, we develop a few indices quantifying various levels of geographical awareness among cities. (1) We find that the geographical awareness proxy by the social media-based indices can measure interactions among cities. Specifically, the geographical awareness among cities follows gravitational law and is highly correlated with mobility flows. (2) The spatial in-awareness index (SIAI) is an appropriate index indicating a city's ranking in the urban hierarchy (3) the spatial out-awareness rate (SOAR) can indicate the interactions from a focal city to other cities. Our findings also show that SOAR can predict the number of people infected during a pandemic in a city system. Once the origin city or hotspots of the outbreak and the number of infected persons within those cities are known, we can use the social media-based SOAR index to predict number of cases for other else cities in the urban system. With this information, governments can properly and efficiently deliver medical equipment and staff to cities where large populations are infected.

Original languageEnglish (US)
Article number103041
StatePublished - Feb 2021
Externally publishedYes

All Science Journal Classification (ASJC) codes

  • Development
  • Sociology and Political Science
  • Urban Studies
  • Tourism, Leisure and Hospitality Management


  • COVID-19
  • China
  • Geographical awareness
  • Social media
  • Spatial in-awareness index
  • Spatial out-awareness rate index


Dive into the research topics of 'Measuring interaction among cities in China: A geographical awareness approach with social media data'. Together they form a unique fingerprint.

Cite this