Spatial data mining has wide applications in different fields, so different types of users are involved. Spatial data mining systems integrate the functionalities of both geographic information systems and data mining systems. They should provide means to manage the complexity of their underlying concepts and to properly assist both experienced and naive users. One possible solution is to enhance them by means of visual languages. Generalizing the concepts of the different spatial data mining tasks, we propose a visual language for spatial data mining named VisMiner, based on the 'Miner Trip' metaphor. It provides a simple and intuitive way to configure all the spatial data mining tasks in a uniform way, to specify the spatial relations to be used by the mining algorithms, and to create different data mining models by exploiting the resolution of precision topological ambiguities.