A Spatially Constrained Infomap for Spatial Community Detection

Infomap is one of the most effective and widely used non‐modularity‐based methods for spatial community detection. However, most studies directly applied Infomap to detect spatial communities without considering the spatial effects, which may generate spatially scattered communities. To address this issue, a spatial version of Infomap called spatially constrained Infomap (ScInfomap) is developed in this study. Specifically, the map equation, an information‐theoretic objective function used by Infomap, was adopted as the objective function for spatial community detection. Additionally, we evaluated the performance of existing optimization algorithms for spatial community detection, including the spatial version of the optimization algorithm in Infomap, focusing on their effectiveness in optimizing the map equation. Our findings revealed that Spatial Tabu Optimization for Community Structure (STOCS) demonstrated the most effective optimization of the map equation. Consequently, we adopted this algorithm as the optimization algorithm in ScInfomap. In addition, experiments on synthetic datasets also demonstrated that the proposed method is superior to modularity‐based methods. A case study conducted using the Xiamen taxi dataset further demonstrated that spatial communities detected by the proposed method are more suitable for urban structure analysis. Moreover, through a brief analysis of spatial communities detected by ScInfomap, we find that these spatial communities can reveal both the uneven development of Xiamen and the geographic constraints.