A Visual Analysis Method for Context-Aware Map Symbol Generation Driven by Large Models

 As a key medium for conveying geospatial information, maps are increasingly emphasizing seman-tic-driven and context-aware representation. Addressing the challenges faced by current map visualization in contextual expression, such as limited context understanding, low controllability, and insufficient inter-active optimization. To address these issues, we propose a context-aware map symbol visual analytics method driven by large models. Based on map context understanding, it parses the geographic spatial ele-ments in user data, extracts semantic information, builds structure masks, and generates map symbols with reasonable geometric structures and semantic contexts that can meet users’ contextualized needs through a latent space diffusion model. On this basis, this paper designs and implements a contextualized map symbol generation visual analysis system, which supports users to input data and perform interactive operations to generate and optimize contextualized map symbols. Finally, through two case studies with different con-textual demands and user experiments, the effectiveness and practicality of the proposed method and sys-tem in context understanding, controllability, and interactive optimization are verified.