Urban landscape simulation in coastal areas based on Geographic Information System and big data

Rapid urbanization in coastal cities requires data-driven approaches for urban landscape planning and spatial decision-making. Geographic Information System (GIS), combined with location-based social network data, help analyze urban activity patterns through user check-in behavior. However, traditional POI recommendation models assume symmetric user influence and often ignore geographic mobility characteristics, limiting their usefulness in spatial planning. To address these issues, this study proposes a GIS-based POI recommendation framework called FUSG (Fusion of User, Social, and Geographic information). The model integrates asymmetric user influence using a PageRank-based global importance measure, social similarity derived from friend relationships and residential distance, and geographic distance decay modeled through a power-law function. These components are embedded into an improved collaborative filtering framework to enhance prediction accuracy. Experiments were conducted on the Gowalla dataset, containing 198,151 users and 6,845,483 check-ins. The proposed FUSG model was compared with four baseline models (US, UG, SGFM, and USG). In Top-5 recommendations, FUSG achieved a precision of 0.13 and a recall of 0.07, outperforming all baselines. The results show that geographic proximity strongly influences urban spatial behavior and demonstrate the effectiveness of the proposed framework for GIS-based coastal urban planning and visitor flow management.