Location-based social media data is, increasingly, an important facilitator of exploring the movement of goods and people in and between countries across the globe. Typical examples include Twitter, Facebook, Foursquare. As with all social media data outputs, the fundamental value of location-based social media data is for sensing users' space-time trajectories, and thus, makes social media data a new platform for understanding business and social interactions in the spatial context. In large developing and emerging economies with massive social media users via computers and mobile phones, real-time 'geo-tagged' human mobility information from social media data sources are clearly potentially large. In these settings, cyberspaces are often built and expanded with the explicit aim of stimulating digital socioeconomic activities and balancing regional disparities. However, despite intense policy and public enthusiasms, there is virtually no direct evidence on exploring the configuration of urban network patterns by using social media users' mobility flows within a large developing country context. The scarcity of empirical evidence is not surprising, given that mining location-based social media data faces serious identification challenges. First, location-based social media data, as a type of big data resource, are often featured by the dynamic, massive information generated by billions of users across space. In truth, despite of the recent development of intensive-computational geographic information system (GIS) modeling programs, social media data with precise individual-level location information is still extremely large to proceed by using the GIS techniques at multiple geographical scales. Furthermore, conventional GIS-based computational methods cannot directly read the unstructured social media datasets (e.g. words, pictures, videos). Additional big data mining methods are often needed to transform social media data information from unstructured data formats to structured, and ready-to-use spatial datasets. In this paper, we tackle these problems by analysing the configuration of intercity connection patterns in China to provide new evidence to the applications of location-based social media data in urban and regional studies. Our examination of changes in human mobility patterns by months by city-pairs throughout China by months involves many potential stages of big data mining analysis. We stratify cities by core-periphery urban systems, by regions and by calendar months, finding that human mobility flows are not distributed evenly over time and across space. We find larger human mobility flows around the Chinese New Year month and the summer months. Our evidence suggests the significantly heterogeneity patterns of core-periphery urban systems as reflected from real-time human mobility flows. As a baseline, this paper is-for the first time in the literature-to comprehensively measure urban network patterns at a detailed spatial degree (the city-pair level) based on location-based social media data from a large developing country context.
