The paper presents a visualization-based approach to explore large databases of geographic metadata. The aim of the data exploration is to search and acquire expressive geographic data to create a geographic information system. Metadata repositories which provide information about available geographic data exist on the web and are usually queried to perform data acquisition task. We propose a visualization–based tool to assist the user in this metadata exploration. The approach is characterized by the integration of multi visualization techniques, interaction methods and data mining concepts. The use of multi visualization techniques and a dynamic interaction with them allow to discover new information among data and assists the user in his reasoning.