Nowadays the importance of collecting and sharing geographical data is rapidly increasing. Spatial data infrastructures are arising to integrate geographic information services which allow to identify and access geographic information from a wide range of sources. The study of data access addresses different issues such as data delivering, data retrieval, data integration and data selection. We focus on the data selection issue: we describe an approach to support data selection activity based on visual analysis of geographic metadata. This facilitates the metadata analysis providing a compact overview of data that are available and a correct interpretation of the result set, allowing to discover properties and relationship among data and to formulate better searching criteria.