Abstract:

This article critically reviewed the development of research perspectives in the studies of regional spatial structure, with an analytical shift from the focus of behavioral aggregation to behavioral interaction, as well as from place-based attributes to network-based interrelationships. Traditional spatial structure research mainly highlights social or spatial correlations among the attributes of intra-regional spatial units (for example, cities), based on the aggregated statistics of social and economic behaviors. Corresponding approach includes social-area analysis and homogenous-area detection. Recent studies have encountered a "relational turn” and "network turn", which aim to frame regional spatial structure into a combination of network structures and spatial patterns from the perspective of behavioral interaction and spatial network. Such network-based analyses include traditional functional-area analysis, community detection algorithms, and core-periphery analysis from the field of network science and social network analysis. By comparing these perspectives, this article finally suggested that future studies should further explore the rationale of behavioral interaction and intra-regional network in understanding regional spatial structure, avoid the trap of "methodological determination" by developing mesoscale analysis, and investigate multiple spatial structures across scales and mobility factors under the perspective of multilayer networks.

Key words: regional spatial structure, behavioral interaction, complex network, community detection, core-periphery, mesoscale structure