<p>This study addresses the growing need to better understand rural tourist behavior from a geospatial perspective under the context of rural revitalization and smart tourism development. By integrating Geographic Information Systems (GIS) with multi-source data, including user-generated content, point-of-interest (POI) data, and spatiotemporal trajectories, a novel analytical framework is proposed to infer tourist behavior patterns and underlying motivations. First, a multi-source motivation modeling approach is developed by combining textual semantics, spatial trajectories, and user attributes to construct comprehensive tourist profiles. Latent Dirichlet Allocation (LDA) is employed to extract latent motivational themes, enabling the identification of diverse behavioral intentions across rural tourism contexts. Second, a GIS-informed spatiotemporal association mining method is introduced by enhancing the FP-Growth algorithm with temporal continuity and spatial proximity constraints. This allows for the extraction of behavior patterns that are consistent with real-world geographic interactions and movement dynamics. Empirical results based on large-scale rural tourism datasets demonstrate that the proposed framework effectively captures geographically meaningful behavioral patterns, significantly improving inference accuracy and rule validity compared to conventional approaches. More importantly, the integration of GIS-based spatial constraints enables interpretable insights into how tourists interact with rural spaces over time. The findings provide valuable implications for destination planning, spatial resource allocation, and personalized service design, contributing to the advancement of GIS-driven tourism analytics and sustainable rural tourism management.</p>